Category: From Youtube

For over a year now, I’ve been getting spammed emails on behalf of this company that calls itself Free Cash, offering to sponsor me if I promote their website and app. Now, when it comes to sponsorships, I always want to make sure that the company I’m promoting has products that I genuinely use and believe in. So, one look at even the company name is all I needed to know that I wasn’t interested. It’s literally called Free Cash. They might as well name themselves Infinite Fortnite V-Bucks glitch 2026 working. Surely anyone with half a brain could instantly tell that a company with this name was up to no good, right? Do not read him that. Do not read him that. Fast forward a few months and Free Cash is still non-stop emailing me. Not only have the agencies they go through been following up emails just a few days without a new response, they had continued to try baiting me by saying things like, “Your competitors are sponsored by Free Cash, too.” This insane dedication of trying to reel me in, started to form an itch at the back of my brain, just what exactly is this company, and how do they have the audacity to email me through agencies this many times? After a while, I decided to play along. And shortly after I was being asked to lie to my audience about how much money people were supposedly making from free cash. That’s when I realized the public information around the company was only the tip of the iceberg and the sponsor is far more predatory than the name alone would suggest. Obviously, there were hundreds of complaints from people that they had not been paid out what they were owed. Not surprising there. But what shocked me the most was how powerful of an entity it was, as well as the size of the creators that have been promoting it. Creators like Tyler Olivea, Darmman, Atosi, WestJet, and the list could go on. The craziest part is that even after being removed from the app store, it looks like they’re secretly making a comeback. Hi, I’m the internet anarchist. I create weekly YouTube documentaries. And today we are uncovering the truth behind free cash, one of YouTube’s largest scams that no one seems to be talking about. Now, I’ve turned down enough free cash money to make my accountant physically ill. If you’re not already subscribed, now would be a pretty good time. So, what actually is free cash? The easiest way to explain is with a story. Tell me in the comments if this sounds familiar. It’s the early 2010s and you’re playing Clash of Clans. Everything’s going great, but you’ve hit that point where every upgrade takes hours to finish. There’s nothing left to do except wait. Then you notice an extra builder hut costs gems. The problem is you don’t have enough gems, and you definitely don’t have a credit card or any money to buy more. In your mind, this is a complete emergency. So, what do you do? You Google how to make free money. Suddenly, you’re flooded with websites and apps promising they’ll pay you to download games, complete offers, or install random apps. They all advertise things like earn hundreds of dollars or make money in your spare time. Maybe you spend a couple hours trying them. Then you realize it’s boring. You make almost nothing and you give up. You go outside, jump on the trampoline, or find something better to do. At its core, that’s basically what free cash is. Imagine taking all of those old make money online apps, combining them all into one giant platform, and turning it into a full-fledged business. It’s almost like what Reddit did to the thousands of small internet forums. Except instead of discussion boards, it’s every online money-making scheme rolled into one place. This business model even has its own name, GPT. No, not chat GPT. In this case, GPT stands for get paid, too. These sites pay people to complete tasks like taking surveys, downloading apps, signing up for offers, watching videos, or testing products. In return, users earn points or cash that can be redeemed for gift cards or real money. The catch is that the payouts are usually tiny. And because this industry has a long history of scams, even descriptions of the business model usually warn people to stick with reputable sites if they want any chance of actually getting paid. Once you start looking into Free Cash itself, things get a little more interesting. Free cash is owned by a company called El Media, a German startup founded by Mora Tolander when he was just 17 years old. The platform wasn’t always called Free Cash either. Before the rebrand in 2021, it was known as Free Skins. As the name suggests, instead of earning cash, users earned CS skins through the same kind of GBT offers. In 2023, the company reorganized under the Al Media name. Looking at their LinkedIn, it appears to be an actual company with real employees, offices, and a growing corporate structure. But some of the numbers they’ve claimed over the years raise a few questions. According to the company, they’ve generated over $100 million in revenue and paid out more than $300 million to users in their lifetime. Independent estimates, though, paint a very different picture. Depending on the source, estimates of their annual revenue are significantly lower than what the company suggests. There’s also another interesting detail. Older review sites claim the same people behind Freeash also operated Gain.gg, another GPT website that’s still online today, although it’s not clear whether it’s still owned by Al Media. So somehow this company claims it bootstrapped itself from a small CS Skin Rewards website into a business worth hundreds of millions of dollars. The obvious question is how? If you ask free cash, the answer is pretty straightforward. They describe themselves as a platform that connects advertisers with users who are willing to complete offers. In other words, companies pay free cash to get people to install apps, play games, sign up for services, or make purchases. That’s all technically true, but it leaves out the most important part. Free cash isn’t paying users out of its own pocket. It’s acting as the middleman. Here’s how it works. An advertiser, usually a mobile game publisher, survey company, online casino, or crypto exchange, pays free cash every time someone completes a specific action. That could be downloading a game, reaching a certain level, subscribing to a service, or making an inapp purchase. Once free cash receives that payment, it keeps a percentage for itself and passes the rest onto the user in the form of coins, where 1,000 coins equals to $1. Those coins can then be redeemed for PayPal, cryptocurrency, gift cards, or other rewards. The more offers people complete, the more money flows through the platform. Every successful offer earns both the advertiser and free cash exactly what they wanted. The advertiser gets a new customer, and free cash gets its commission. That’s the part most people miss. Free cash isn’t really selling free cash, it’s selling users. The product isn’t the rewards people receive. The product is access to millions of people who are willing to download apps, answer surveys, and spend money because they’re being promised a payout. The cash is simply the incentive that keeps the entire system running. Now that we’ve looked at Free Cash’s background and how the company says it operates, it’s time to look at what everyone else says actually happens. The biggest complaint you’ll find, no matter where you look, is that people don’t receive the money they were promised. Normally, you could dismiss that as the internet being the internet. Every company has angry customers, but what makes this different is how consistent the complaints are. The same story has been told by thousands of unrelated users across multiple countries on Reddit, the Better Business Bureau, Trust Pilot, and countless other forums. These reports stretch back over 5 years, and despite coming from people who have never interacted with each other, they all follow almost the exact same pattern. First, everything works perfectly. New users complete a few small offers, request a payout, and often receive their money without any issues. That builds confidence and makes the platform seem legitimate. Once users begin completing larger offers and accumulate a more meaningful balance, something suddenly changes. Their withdrawal is paused, their account is flagged for a security review, and before long, they’re permanently banned. The explanation is almost always the same. Users are told they violated the terms of service, used a technical exploit, or engaged in suspicious activity. When they ask what they supposedly did wrong, they’re almost never given any evidence. Instead, they receive the same generic response, explaining that the security team reviewed the account, found violations in their internal systems, and that the decision is final. What’s interesting is that Freeash openly acknowledges part of this process on its own support pages, stating that when suspicious activity is detected, cash outs are paused while the account is reviewed. The controversy isn’t that reviews happen, it’s that many users claim those reviews end with permanent bans while providing no explanation beyond a template response. The Better Business Bureau paints a similar picture. Over the past several years, hundreds of complaints have been filed against the company, with many of them describing nearly identical experiences. A common theme is that users are able to cash out small amounts without any trouble, only to lose access to their accounts after earning a larger balance. One complaint from 2026 describes a user who had accumulated more than $200 before suddenly being accused of manipulating the platform through telemetry anomalies. According to the complaint, the account was banned shortly after the user completed a higher paying game offer. The user argued they didn’t even own a computer capable of running the software they were accused of using. Whether that individual account is actually legitimate is impossible to verify from the outside, but the wording of the company’s response looked nearly identical to countless others. The same discussion plays out constantly on Reddit. The official free cash subreddit has dedicated threads specifically for banned accounts and paused cash outs because the volume of posts became large enough to organize into a single place. Another source of frustration comes from the offers themselves. Many of them require users to spend real money before qualifying for a reward. People have reported depositing money into online casinos, purchasing in-game bundles, or subscribing to services because the offer promised a larger payout than the amount they spent. Then after completing every requirement, the reward was either never tracked or was rejected entirely. Whether those failures were caused by the advertiser, tracking errors, or free cash itself often isn’t clear to the user. The end result is the same. They spent real money expecting to be reimbursed and instead they received nothing. Notice how I used sources like the Better Business Bureau and Reddit to understand what people are actually saying instead of relying on the usual review sites you find when you search free cash. One of the most common ones is Trust Pilot, which you’ve probably come across before. The reason I did that is because I am almost certain those reviews do not reflect the full picture. When you look at Trustpilot, the ratings are sitting at a nearly perfect five stars, which immediately feels unusual for a platform in this space. Trust Pilot in general is known for being relatively easy to game, especially in industries where companies have a strong incentive to manage their public image. More importantly, when you compare those ratings to what you see across other platforms, the story changes completely. on Reddit, on the Better Business Bureau, and on independent forums. The feedback is overwhelmingly negative, or at least heavily mixed with far more people reporting issues than smooth experiences. That contrast is what makes the Trust Pilot situation hard to ignore. It creates a kind of noise in the data where the positive reviews exist, but they are surrounded by a much larger volume of complaints elsewhere. And once that happens, it becomes difficult to treat any of the highly positive reviews as fully reliable on their own. The end result is that it becomes almost impossible to tell which experiences are genuinely representative. Any real positive feedback gets lost in the suspicion that the review system itself may have been manipulated, especially in an industry where users are already skeptical to begin with. Despite this, it’s still obviously impossible to verify, even if thousands of people are pretty much saying the exact same thing. And another YouTuber already figured this out and decided to go into the app himself to try seeing if he could earn any money. Around a month ago, YouTuber The Asher Show released the only other video I know about free cash in which he describes his experience of trying to use the app for a month. Almost immediately, the app started asking surprisingly personal questions about his habits, purchases, and other information before finally pushing him towards its offers. The question it was asking me if I’ve done things like inapp purchases in mobile games recently, and has asked my name. I’m surprised they didn’t ask my mother’s maiden name and my blood type. And there’s a reason that it was asking all these questions. Right after signing up, the app rewarded him with a collection of bonuses just for creating the account. His balance climbed quickly without much effort, making it feel like he was already on his way to earning real money. The first game Free Cash recommended was a simple mobile game where he spent a few minutes tapping through levels and collecting rewards. After completing the first milestone, he earned only a few cents. But because of the signup bonuses and streak rewards, his account balance looked much larger than the amount he had actually earned by playing. It was enough to convince him to keep going. That confidence disappeared as soon as he tried to cash out. He discovered there was a minimum withdrawal requirement that he hadn’t yet reached. And that minimum is $25. The amount is different depending on where you are. I think it’s $16 in the UK, $25 in the US. The app made it feel like he was close enough that continuing to play would eventually get him there. After finishing the first game, Free Cash unlocked even more offers. Many of them advertised enormous payouts, sometimes claiming players could earn hundreds or even thousands of dollars. Looking closer, Asher realized those rewards were divided into dozens of separate milestones. Each step required more time than the last, and the biggest payouts were hidden behind extremely difficult objectives. Then the requirements changed. One offer promised a large PayPal bonus, but only after reaching milestones that became increasingly unrealistic. After spending hours playing, he realized he wasn’t even close. Another challenge claimed he could earn over $20 for reaching a certain scene in the game. Then, an even larger reward for making it several chapters farther. At first, those goals sounded achievable. After nearly a week of playing, he finally understood they weren’t. Six straight days of progress had only brought him to scene 4, while the reward required him to reach scene 9. Even after more than 100 levels, he wasn’t even halfway there. I’ve been playing this game for 6 days. I’m on level 103 and I’m only on scene 4. It’s taken 6 days for me to get to scene 4 and it wants me to get to scene 9 just to make 86 bucks. It’s not happening. The app kept presenting these rewards as if they were just around the corner. But every big milestone revealed another much larger one waiting behind it. By the end of 7 days, after spending hours every day completing challenges and playing games, Asher had earned just over $12. After another day, his balance finally reached a little over $15, but it wasn’t enough to cash out. The closer he got, the slower the rewards became. Looking back on the experiment, Asher realized most of his balance had come from the promotional bonuses he received at the very beginning, not from actually playing the games. By the end of the test, he concluded that the experience never matched the way the app was advertised. The large rewards were technically possible, but only if someone could complete an exhausting series of increasingly unrealistic milestones. After more than a week of testing, he still couldn’t withdraw a single dollar. Advertising is a massive reason why Freeash has the reputation it does today. In fact, the marketing became so deceptive that Tik Tok eventually removed its ads from the platform for violating its policies. The biggest selling point wasn’t the app itself. It was the promise that anyone could get paid simply by scrolling through Tik Tok. Ads claimed users could earn money just by watching videos, making it sound like one of the easiest side hustles imaginable. Some even suggested people were being hired to scroll social media and could make $35 an hour just by watching Tik Tok. Obviously, the reality was completely different. The entire marketing campaign was built around an experience the app simply didn’t offer. As those ads spread across Tik Tok, Free Cash exploded in popularity and climbed to the top of the app store charts. But once Tik Tok reviewed the advertisements, it determined they were misleading because they falsely represented how users actually earned money. The platform removed the ads for violating its policies on deceptive financial claims. Free Cash later claimed the advertisements had been created by affiliate marketers rather than the company itself. But regardless of who made them, the campaign had already done its job. Millions of people downloaded the app expecting to get paid for scrolling Tik Tok, only to discover that wasn’t what Free Cash actually offered. In January of 2026, a Wired article would get written about this topic, as well as a Malware Bites article. For those of you that don’t know, Malware Bytes is a cyber security software made to block viruses. But more importantly, they have a vested interest in talking about consumer rights to privacy. In the article, they say that our concern beyond the bait and switch is the privacy issue. Frecash’s privacy policy allows the automatic collection of highly sensitive information, including data about race, religion, sex life, sexual orientation, health, and biometrics. Each additional mobile game you install to chase rewards adds its own privacy policy, tracking, and telemetry. Together, they greatly increase how much behavioral data these companies can harvest about a user. Experts warn that data brokers already trade lists of people likely to be more susceptible to scams or compulsive online behavior. Profiles that apps like this can help refine. We’ve previously reported on data brokers that used games and apps to build massive databases only to later suffer breaches exposing all that data. A few months later, a TechCrunch article would site these two articles as well as provide their own bombshell. Not only was Free Cash’s metrics most likely botted and fake, no surprise they were literally in between chat GBT and Claude with most used apps, but it seemed as if they had created or bought another company as a burner in order to evade a ban. reading third-party data from market intelligence provider app figures shows that Alia GmbH, the German company behind the Freeash app, first submitted free cashache to the App Store on March 24th, 2024, but it was removed after around 2 months on June 13th, 2024. This version of the app was downloaded an estimated 69.5K times before removal. Months later, a second app that already existed on the store, developed by the Cypress-based company 256 Rewards Ltd, was rebranded as Free Cash, and an update was submitted under that app’s ID. According to the third party data, it’s unclear if L Media acquired the Cypress-based developer, which was originally called Rewards, or if it simply acquired the developer account. Rewards no longer appears to have a functioning website or social media presence, and the team behind the company has since moved on to a new company. Pushed Alia’s re-entry into the app store through another developer account may have been a way of circumventing a ban on the initial free cash app. Using another developer to re-enter the app store after a ban is a common though rule-breaking tactic. Al Media’s spokesperson declined to comment about its earlier app takedown. After reaching out to Apple for comment, Free Cash was promptly taken down from the App Store as well as the Google Play Store a little while after. Despite getting banned from advertising on Tik Tok, Freeash is still free to advertise on just about every other platform, especially YouTube. And the number of creators who have accepted sponsorships from them is honestly pretty shocking. Like I mentioned earlier, creators like Tyler Olivea, Darmman, Atosi, Legal Mindset, Cinema Sins, and plenty of others have all promoted Free Cash. Some of these channels have millions of subscribers and several videos promoting it. Again, I’m not saying these creators were secretly working with Freeash or knew the company was shady, but come on, the company is literally called Freeash. A 5-minut Google search brings up pages of controversy, complaints, and accusations that it’s a scam. Even a quick look into how the platform works should raise a few red flags. If you look through the comment sections on these sponsored videos, if they aren’t deleted, you’ll almost always find people warning that it’s a scam. In some cases, the creators even replied defending the company. Again, I’m not trying to call anyone out or claim they knowingly promoted something malicious, but I do think it’s worth pointing out just how questionable it is to take a sponsorship from a company with a reputation like this. That being said, there’s one more piece to how freeash gets creators on board. All of that outreach I mentioned before came from agencies working on their behalf, not free cash directly. And some of these agencies use a referral model where one creator refers another creator to promote it and then takes a percentage of everything that the second creator earns from it. I haven’t fully pulled this thread yet. So, I can’t tell you exactly how deep it runs. But once you have people earning a cut of whoever they bring in, the whole thing starts to take the shape of a pyramid while also giving creators who were told by other creators to promote it a false sense of security as it leverages the existing trust and relationship. And to be fair, on the surface, the deal looked quite appetizing. As I’ll get into later, Free Cash is paying big time. And in some cases, when I pushed back against promoting it, they pointed to my competitors and to videos with hundreds of thousands of views already promoting free cash, trying to make it feel like this is just something everyone’s doing, so it’s probably fine. Now, I don’t think these creators are malicious. I think a lot of them were careless and everyone is careless at some point, especially with thousands of dollars sitting right in front of them and every initial signal telling them it’s safe and in many cases other creators telling them how awesome it is to promote or how much money they’ve been making. I’m not here to stand on a moral high ground. Every creator, me included, makes mistakes over a long enough run, but it’s not hard to assume that they were being paid very well to overlook those concerns. Which finally brings us back to the reason I made this video in the first place, the emails they sent me. Now, we can actually look behind the curtain and see what Free Cash is offering creators to promote the platform. Out of the dozens of emails they sent trying to get me to accept a sponsorship, one of them offered me an $8 CPM. For anyone who doesn’t know, CPM essentially translates to cost per,000 views. So, that’s $8 for every,000 views your sponsored video gets. If a video reaches 100,000 views, that’s around $800 from the CPM alone, which is already better than what YouTube typically pays through ad revenue. Aside from the CPM deal, creators can also earn money from referral signups. Every time someone downloads free cash through their link and completes the required actions, the creator gets paid. To match the CPM alone, they would only need around one sign up for every thousand viewers. That’s about a 0.1% conversion rate. In reality, campaigns like this often convert around 1% or even higher. In other words, creators promoting free cash can make an absolute fortune. While free cash is spending huge amounts of money trying to get as many people onto the platform as possible. And when you think about who these ads are aimed at, it starts to make a lot more sense. They’re marketed towards people looking for quick money, many of them younger viewers or people in difficult financial situations. It’s the perfect sponsorship opportunity for creators willing to cash in on an audience that’s already vulnerable or to trick creators who are convinced by agencies of otherwise. During one of the email exchanges, I decided to play along and act like I was interested. Eventually, they sent me a template script they wanted me to read in my video. Now, sponsored scripts aren’t unusual. Pretty much every brand sends creators talking points, but one section immediately stood out. Yo, what’s good, y’all? I’ve been putting people onto this program called Free Cash for a while now, and my DMs keep blowing up with people saying, “I tried it and just cashed out.” I love seeing that because it shows this really works. Now, I’ve seen a few comments from people saying stuff like, “It doesn’t work or they just want your data.” That’s not true at all. Freeash doesn’t sell your data. They only track your progress so they can make sure you get paid. You literally just download and sign up with an email and password. No personal info needed. This was one of the biggest reasons I decided to make this video. They were literally asking me to lie to my audience. At that point, I had never promoted Free Cash. I had never worked with them, and nobody had been messaging me saying how they had successfully cashed out. Yet, the script wanted me to pretend my DMs were full of happy users. It also wanted me to pretend that I had already seen comments calling free cash a scam just so I could immediately dismiss those concerns on the company’s behalf. Think about that for a second. They get accused of being a scam so often that they built a response into the sponsorship script before anyone even leaves those comments. It’s honestly hilarious. And more than anything, it shows exactly who these ads are trying to reach. They’re aimed at people who need money quickly. people who might not stop to question the claims they’re hearing from creators they trust. And that’s what makes this marketing so predatory. At the end of the day, the way Free Cash is set up is honestly kind of genius. It’s not a rugpull. It’s not some wallet drainer that steals your credit card information and disappears overnight. It’s a slow bleed. The platform doesn’t have to scam everyone to be successful. In fact, it can’t. And that’s what makes it so effective. Instead of being a simple nobody gets their money scam, it’s something even more slippery. It’s a real ad tech business that pays small amounts to many users while showing a large consistent multi-year pattern of denying the larger payouts marketed through provably false advertising and a paid affiliate funnel that makes its YouTube reviews anything but independent. Seriously, try watching any free cash review without a free cash affiliate link in the top of the description. As I was writing this script, somehow Free Cash has even made its way back onto both the Apple App Store and the Google Play Store. How they managed to get relisted is unclear, but for now, they’re right back where they started, ready to attract another wave of users. On the bright side, looking through the hundreds of complaints on Reddit has revealed a certain user that says they are apparently starting a class action lawsuit against Freecash. Whether or not this is actually real or not is up for debate, but there’s no public legal filings available to see as of writing this. Whilst editing this video, I went down a rabbit hole and accidentally discovered how Free Cash is secretly controlling the narrative. One interesting note is that the official Free Cash subreddit is actually moderated by Freeash themselves, which explains why so many posts complaining about Free Cash’s support or accounts being frozen are promptly removed from the sub. And even making a post mentioning that you’ve been muted or banned for talking about the support will get you permanently banned as well. If not permanently, then for a week. The next part is even worse, and it explains how their Trust Pilot ratings are so high. As I have mentioned in this video, it’s really strange how when you look on Trust Pilot, they have an almost perfect score, but you look anywhere else with an independent review and suddenly it’s just people mentioning how they weren’t being paid out, how their accounts were frozen, and how they weren’t getting any help from support. Under this Reddit post here of someone explaining their story of them making a review and the review being taken down, I learned something about Trust Pilot, which explains why Free Cash has such a good score. So basically this post is a user explaining how they were using free cash and they were being accepted for their lowpaying task but the moment they tried a higher paying task the tracking for the offer suspiciously stopped working when they completed the offer. Now this user went to Trustpilot to make a review explaining their experience to try and warn others of what might happen to them if they try using free cashache. Upon making their review, Freeash disagreed with it and actually flagged their review, meaning the user had 72 hours to provide evidence or the reviewer to be taken down. Now, this immediately stood out to me because as I started looking deeper, there were tons of other users who tried leaving negative reviews only for Freeash to claim that the user never used the website and that the review was inauthentic. And this basically explained why Freeash has such a positive reputation on Trust Pilot but not really anywhere else. Looking deeper into Trust Pilot and they actually reveal the data around the reviews they flag. And what I found here is kind of insane. So basically if you look at Free Cash’s flagging activity in the past 12 months alone, they have flagged almost 700 reviews with almost 80% of them being one-star reviews. As I touched on before, once a review gets flagged, you have 72 hours to respond with evidence or your review gets taken down. And as this user here has mentioned, reception 16685, they’re basically counting on you not noticing or not bothering to send the evidence. And clearly, it works. As the stat here says, less than 1% of the flagged reviews were flagged for invalid reasons. Basically meaning that 99% of people weren’t able to appeal their review and therefore their review was taken down. Out of these hundreds of reviews, only 31 actually remained on Trustpilot and 665 were removed. And this is just in the last 12 months. And if you thought that this was already bad, it gets even worse because when you actually look at the sources of where these reviews are coming from, most of these reviews are what’s called invited reviews. So, it’s when Free Cash is actually sending the user to directly review the site, not just when a user decides they want to leave a review based on their own experiences. Now, obviously, we can both put two and two together. If there’s an app that’s offering to give you money and has a reward system and they ask you to review it, obviously the user is going to be incentivized to leave a positive review, especially when that review was invited by Freeash. Now, I can’t confirm this for myself, but another Reddit user did comment that the Freeash head of support used to be the global director of quality and compliance at Trust Pilot, which if true would explain how they’re so good at gaming the system. Not that you really would need to be because the deeper I look into Trust Pilot, the more I’m sort of learning that Trust Pilot itself is basically a front for sketchy businesses to appear trustworthy. And in my opinion, Trust Pilot should never be used as an actual data point when deciding whether something is actually a scam or worthwhile. One thing I do want to note here is that the deeper I dug into Free Cash, the more I feel completely unqualified to talk about it. The whole thing is just way bigger than I expected going into it. And I feel like someone like Coffeezilla or someone who’s actually professionally trained in investigating this kind of thing, needs to look into it because with just what I found already, there seems to be a lot more to this. And in my opinion, it’s bordering massively on fraud. By the way, one of the ways you can cash out your earnings is by transferring them directly into a stake account. And I would bet that mysteriously has a better chance of working. That one detail feels like the perfect summary of where YouTube sponsorships have ended up. Some of you probably remember the controversy surrounding T. Martin and Pro Syndicate. Years ago, they were exposed for promoting a CS gambling website without telling users they actually owned it. At the time, it was one of the biggest scandals YouTube had ever seen. Now, compare that to today. Creators regularly have undisclosed stake ads all the time. And to be clear, I’m not making this video to call out every creator who’s accepted a free cash sponsorship. I don’t think most of them were intentionally trying to mislead their audience. What I’m trying to say is this. First, I don’t think free cash should be anywhere near the app store, let alone paying YouTubers to promote it. And second, I think it’s time we raise the bar for what we’re willing to advertise. Whether it’s free cash, stake, or the next company promising easy money, these businesses all rely on selling the same fantasy, quick cash with little effort, more often than not, the people buying into that fantasy are the ones who can least afford to. It dawned on me that Free Cash might actually want to respond to this video, and I was thinking what they would probably say. I believe it would be based on what they said after they were banned from advertising on Tik Tok due to the false advertising claiming that people could scroll and earn money. In which, as I mentioned earlier, they simply responded by saying that those advertisements were made by third parties using their affiliate system and that in the future they would try harder to moderate them. In my opinion, this is simply just an easy out. And the fact that I’ve been reached out to by so many different agencies and seen the same scummy sales tactics trying to get me to post a free cash ad tells me that this isn’t just one agency that’s gone rogue. It’s at worst a partnership with Free Cash to just get as many users as possible, no matter how deceptive the tactics are, or at best freeash completely failing to audit and moderate the advertisements of their app. Either way, if this video could have any impact, the impact I hope it has is that it serves as a resource for creators who are considering taking a free cash ad and who are seeing the money being waved in front of them or perhaps are almost falling victim to the same tactics they tried on me to get me to promote it. Throughout this video, when I say they, I am referring to all of these different agencies representing Freeash. I do want to make clear that I have not had any direct contact with Free Cash themselves and I’m not even sure if outside of their affiliate program and their partnership with agencies that they are directly doing deals with creators. It’s

Motion designers are dead according to the new VIP motion tool from Hickfield AI. But let me tell you a little secret. First of all, I’m not sponsored by Hickfield, otherwise I would also not say this. But what I’ve seen is that a lot of these videos are just Invado templates made in After Effects. But still, there was something in me that was thinking maybe maybe there is a way to get a good result out of this Hicksfield Vibe Motion tool. And that’s why I want to do a battle a battle between Hicksfield and After Effects. So, place your votes in the comments. Who’s going to win? AI Tom or Adobe Tom. Let’s start this battle. All right, let’s make some jobs redundant, shall we? More AI, less employees, more profits. I’m here in Hicksfield, and what’s interesting is I couldn’t find it on the top bar, but just had to go to Hickfield Motion. Maybe it’s because it’s in beta. I don’t know. Let’s click on from scratch. I want to do it from scratch. I think that’s only fair. Now, let’s go to Adobe Tom. All right, let’s speed run this new composition. Okay, let’s create a new solid. And let’s add a grit to that blending mode. Normal width. Make it like nice and big. Color bit darker. Something like that. Then I’m going to key frame the width. Going to make it a bit further. And let’s zoom it in. Text fully stack editor club. Center. Center that. Then make that nice and big. And then let’s change the color to like almost black. And I’m going to use the bouncy word preset in my plug-in. If you want the plugin, it’s in the fullstack editor club. Link is down below. Back to you, AI Tom. Now, I’m going to first keep it simple. A simple logo animation. So, I want a simple logo animation where the words full stack [music] editor club fly in from the bottom just like where all the editors should be at the bottom. white background, black text, a small orange accent color, light grid background. Let’s just keep it like this. I’m just going to click generate and let’s see how long this takes. Let’s get back to you, Adobe Tom. Thank you, AI thumb. Now, let’s set the bounce length to four and the bounce maybe to 40. So, we just have this clean animation. Nothing too special, nothing too crazy. I’m going to add a line pencil. Whoop. Let’s click here, hold shift, click there. And let’s add a stroke to this. And let’s make that wrong. Just like that. Bit thicker. Maybe 10. 10 pixels. Or is that too thin? Maybe a bit too thin. Let’s move it a bit more down. And let’s add a trim path to this. Quick key frame on the end. Start 50. And the easy ease on that motion blur. And then I’m already going to select these layers. And then add [music] a new null to this. And then I’m going to do something that AI Tom can’t do, and that is create a cool transition. Easy ease that thing. Very nice. Next scene. Back to you, AI Tom. Yeah. So, that’s when I fired 500,000 employees, and I got away with it. I’ve been thinking of this AI ID. So, what if we could replace food with AI? Yeah. And then maybe we get funding for it and then we go bankrupt. I think that’s a good idea. Yeah. It’s done. Wait a sec. Oh, I’ll call you back. All right. So, we have a 2C animation. Okay. We have a 2C animation. We have an animation. I mean, let’s be honest. It’s not bad. It’s not bad. It’s It’s something, I guess. Yeah. Yeah. It’s something. Now, a quick break in this battle because there is a AI tool I use every single day. And this is one of those tools that doesn’t replace you as an editor that also doesn’t pretend to be a replacement of editors or animators. A tool that is there for editors and for creators. Now, that tool is Submagic, and I use it for all my captions on my short form videos, the ones I produce for my clients, but also the ones I produce for myself. It just saves a lot of time. And what’s cool is that you can customize it. But there’s even some AI tools that are really handy. For example, audio announcement, but even some basic editing. Now, you’re still the director in the director’s chair. You choose what you want in this short form video, which clip is the best, which caption to use, but it saves so much time. It’s also way more accurate in terms of the subtitling itself. So, you just save a lot of time uh changing captions around. They even provided me with a discount. So, link is in the description. And then, let’s get back to the video. Let’s create a new animation. And this time, I’m going to use a card. And I want to create a credit card animation. [music] Make an animation with a dark background with a soft orange blob with a luxurious card credit card with the text make more money with editing and a what is it called? Total money counter going up. By the way, estimated cost 8 to 60 credits. That is quite quite a big range. I’m really curious how much credits we spent at the end of this. As a AI bro, I of course don’t care, but I do really care.

I might make some coffee from here. I’m going to create a new solid. There we go. Let’s add a cool background to this. I’m going to just use the ellipse. Let’s just add a solid color. Very orange. Going to add a quick radial blur to this. A quick wiggle path. Size huge. Huge. And now it should move around and bubble and wobble. Quick CC reptile on this. Look at that lava lamp effect. I love it. I’m going to cut that in. Let’s just have a hard cut. Now a quick credit card. Rounded rectangle tool. Add like a darker blueish color to this. Probably something like this. Center. Center. CC light sweep. This can be a bit darker. Stroke. [music] Same color. Just a tiny bit brighter. Nice quick key frame. There we go. Looks clean. That’s a clean card. Let me quickly add some elements to this Mastercard logo. Click on the fill linear gradient. See if we can make this uh gold goldish around here. That works. Very good. Very nice. Now, let me add some credit card numbers. And for that, let me just grab my credit card. No, let me just go to the ellipse. This color. Let’s make [music] it smaller. Just going to add a quick repeater to this. It should work like something like this. I want four. And then I want to have this again. Nice. Maybe a bit lighter. Text tool to or just balance. Balance. I want that white. And I [music] want that regular and smaller. That’ll do. Nice. Going to duplicate that currency. Currency counter monus base. Let’s animate this stuff. Let’s start at 6K. Align that to the left. Set a key frame. Drag it out a bit. Great. I like it. Made a bit more dark. Bit more luxurious. Luxurious. Now I want to have a card behind that. I think that’s a cool design where it’s a bit like stacked almost. See, I can make these adjustments [music] and AI Tom can’t. But before I do, let’s get back to AI Tom. I’m back. I’m back. I vibe coded 50 websites in the meantime. What did you do? Let’s see which animation or what the animation is looking like. Okay. Okay. Okay. Okay. That’s cool. But I know you can also change details. So, let me change some details here. Let’s see in the background if we can change the color here. Okay, we can’t. Wow, amazing. This works so well. Make more money with editing. Does this work? Oh, I broke it. Okay, you can’t add an enter after it anymore. So, let’s just keep it like this. Let’s not change it. Size. Okay, that works. That works. Color also works. I wish I could add a gradient to this. I don’t think I can. Uh, ease in, ease out. Does that change the animation? Let’s see. Linear. I don’t know about you. I don’t see the difference. I I there I don’t think there is a I think this is just for fun. You can just press this button if you want to. If you don’t have anything to do and the fund weight, that also works well, I see. Okay. And the money counter. Let’s see what can [music] we change here. Start amount. End amount. Let’s change the end amount. That doesn’t work. And the start amount also doesn’t work. And the color I also think that probably doesn’t work. No. No, [music] no. Great. Cool. I wonder how Adobe Tom is doing. Make more money with editing. Let me align this in the center. Bounce, bounce, bounce. Let’s set this to 30. Now, of course, uh this needs to uh animate. And because with the previous one, we had a motion that goes up, we can do a match cut where this also moves from the bottom to the top, [music] make it drop. Let’s just select all these link this to this card. And we’ll call this main card. Set some motion blur to this. Maybe also tiny bit of a scale. And also an opacity. And same goes for the other elements. Something like this. I want a bounce on the position scale. And then the other ones uh that are in the back. I don’t want to have them appear yet. These are these two. So I’m just going to key frame the scale, the opacity, and the position. And then let’s drag these out. Let’s turn the opacity off. Very nice. So they appear in the background. And of course, a nice ease. Make it ease. I like it eased and smooth. Oh, [music] that lava lamp is going well. Back to you, AI. Thanks. Next animation. I want a white background with six [music] blocks that act as classrooms. I want the text unlock your editing potential. V in. Let’s keep it Let’s keep it more broad. animate in and then join the school. Let’s hit generate. I want to see how you did it, Adobe Tom. I’m glad I can finally get back to it. Thanks, AI Tom. I’m going to morph this into our next scene. Now, I want to link all of the elements that are not linked. I’m going to just link those quickly to my null controller, which means I can grab that null controller and create a new null. Press S for scale. And I’m going to just have a slow zoom on this cuz I want [music] that continuous motion. And we can even use this null for a transition later on. For now, I’m going to grab the main card. I’m going to go into the contents because I want to change the size. [music] And instead of icons, I want to just change it to a classroom object like this. Now, let me change the color of this first. I don’t like it. Troneeone. [music] That could work. We then maybe grab a color that’s near here. That will look cool. Nice. Now, it [music] is already a bit in that shape. Probably around here. Also going to set a key frame for the roundness. Now to move it, let’s just grab that null controller. Set a key frame. Probably something like this. I want the bounce on that. A nasty bounce. Make it ramp in. And when this happens, I want these elements to all go away. So trackmat them. Turn it back on. Let’s set a quick key frame. This can go to the right. I’m going to fade out all the rest of the elements. And then lastly, a new white solid. Very nice. So, we have a nice match cut. Um, a nice smooth transition. Back to you, AI. Tom, productivity comes from all the things that you choose not to do. I see. [sighs] Why is this taking so long? What is this? Like, are we contacting aliens? That is the question. What are we thinking about? I could have vibe coded this. Why does this take so long?

I probably could have animated this around six times already. Now we replace 500 employees with just one AI agent. If you want to market the agency, find your employees right now. Let me explain. It’s done. It’s done. It’s finally done. Finally. I didn’t expect this day would have come. Let’s see what it looks like. I’m curious. Let [laughter] me Let me get my binoculars. What is going on here? Can I change this? Why is it so small? That’s what she said. It’s something I never heard. No, but really like why is it so small and I can’t I can’t adjust it. No, I can’t adjust it. Well, this is just uh how it is then. Third animation. And um that’s just uh yeah, that’s [laughter] that’s just it, I guess. How is the animation? Yeah. Okay. Okay. Let’s go to Adobe Tom. Script that element that we had. Oh, almost perfect. Let’s link it. Trackmat it. Cool. Now, I just need a few more elements, more classrooms and a text and a button. So, I duplicated the shape. remove all these key frames that I don’t need. There we go. Something like this. So, we have this one pop up. And then the rest I want to also do the same. So, now I can drag in another classroom, add that same tronee to that, link it, and trackmat it and scale it down. And I will do that four more times. There we go. Now, I need to just replace them. So, hold alt ploop. There we go. Very nice. Just some offset. Now, this is at least readable, right? Let’s add the text to this. Nice. Maybe a soft white gradient on this. I’m just going to fade that in. Nice text bouncy word. And again, continuous motion. We can grab that null controller. And I’m just going to already link this again. And this is also the cool thing. We have control over where the eye goes, right? Like now it goes a bit more to the top. And then you’re like, what is happening here? We’ll move the text with it later on. First, I’m going to easy ease this. And then we can add a cool button here. Rounded rectangle tool. And I can change the color of this. So we can always change it later on. Something like this. Full roundness. Add a quick text to this. Join the club now in white [music] in medium and a bit smaller. Link it. Quick scale. Trackm opacity key frame floop and a [snorts] fade word. And because this now moves up, your eye automatically goes to this button where it says join [music] the club now. All right. Thanks AI Tom and thanks Adobe Tom. We’re going to of course vote on both. So again, do leave the winner in the comments. I’m going to look at them and I’m going to judge on design, motion, and speed. So let’s start with the AI one. I think actually the design is not bad. A lot of things are quite nice in what they look like and a lot of beginner editors struggle with this. So I think I it would actually be a cool tool to use for some inspiration maybe. Now, with the design of the After Effects one, I think there are some improvements here and there, but contextually, I do think it’s better, but the padding and stuff like that, the AI did win in my opinion. So, let’s give this one to the AI. Now, let’s go through the motion. And let’s be honest, the motion is really, really boring with the AI one. You can clearly see that this is a JSON CSS type of animation where it’s really simple. There’s a ease in and ease out. There’s no overlapping motion. And it’s really boring in that sense. Everything fades in at the same time. There’s no different elements that pop up. There’s also no psychology behind it. It’s not like, okay, let’s pop up this first. So, you think of this. Lastly, there’s zero transitions between it. It’s really like just hard cut. And although this is a beta and this is a really early version, maybe there will be a better version of this. However, what I do think is that this is really built on code. So, it’s going to be a really big challenge to have it morph between things, have smooth motion between it, have some motion blur in it. So, this round 100% to Adobe top. The last one, speed. And as you might have noticed, the AI tool is really, really slow. And maybe this will improve in the future, but maybe it also won’t, and it will just be slower once more people are going to use it. You would say that it would go quicker with all these server improvements and stuff like that. But what we’ve also seen before is that once more users use it, it just becomes slow as as crap. Now, you can do something in the meantime, which is nice. However, you have to wait constantly on a end result, which also means that sometimes you did a prompt, it’s really bad, the end result, and then you have to start it again, prompt it, and maybe it’s again really bad. Now, they say that you can adjust things and stuff like that, but you’ve noticed this too is that it’s not there. You can’t adjust the text. You can’t adjust the motion. And then what really frustrates me is that these end results that we see on Twitter or on Instagram, they’re either not made with this tool at all or if they are made with the tool, it’s prompting for days and days and days. And that costs a lot of time, but also takes a lot of credits out of your account. Because aside of the speed, I probably would go through all of my credits uh if I just prompt a few times or if I have to reprompt a lot of times. So, with speed, I would also say Adobe Tom is the winner. Now, will we ever get to a point where After Effects wouldn’t win? Maybe. I don’t know. We’ll see. It is cool to try these tools out cuz maybe you can already incorporate it somehow in your workflow. And then being an expert in After Effects will help a lot because you will know these terms like ease out, ease in, what that means. But also just making something look

Hermes Agent is the number one AI tool you should be learning right now. It is a super powerful AI agent that is basically your 247 AI employee. The problem is most people have absolutely no idea how to use it or get any sort of value out of it. In this video, I will show you how to use this incredible tool. I will cover literally everything from installation to get it up and running and choosing the right model, choosing the right computer to use cases, everything. If you stick with me until the end of this video, you will be a master of the most powerful AI agent on planet Earth. Now, let’s lock in and get into it. If you haven’t used or mastered Hermes Agent yet, this is about to be the best 60 so minutes of your entire life because it is going to change everything for you. This is Hermes Agent. What I’m going to show you in a second. First is what it’s all about, everything you need to know about it. Then I’m going to show you how to install it, use cases, all that. This video is built to be jumped around if appropriate. So, if you don’t need to see installation, if you already know what Hermes Agent is, you can look down below and find different sections. So, feel free to jump around however you’d like. But, let’s get into it. First, let’s first talk about what Hermes Agent is for those that are new or for those that just need a refresher. After that, we’ll talk about Hermes versus OpenClaw, which one you should be using. We’ll talk about which models you should be using. We’ll talk about should you use local models, should you put this on a Mac Mini. We’ll talk about where you should install this, how to use their dashboard that I promise you 99% of people are using wrong, their built-in automation and cananband board, how to take advantage of their self-improvement. We’ll talk about which skills you should be using, use cases, security, all of that. So, jump around for that. But first, let’s start off with what Hermes agent is. It is a 247 AI agent. That means it can work for you around the clock, completely autonomously, completely proactively figure out tasks and activities you should do to advance you, to make you more money, to get you closer to your goals. It is a custom employee for you. And I truly believe it’s the most powerful AI technology out there right now. It learns everything about you and the way you work. It learns your goals, what activities you do every day, what tasks you do every day, how you make money, what you’re trying to achieve. It learns everything about you and saves it into its ultra powerful memory. The most powerful memory in AI agents. It’s accessible from everywhere. So, you don’t have to install any sort of Hermes app. You can just have this integrated into your Telegram, your Discord, your iMessage, whatever you want. So, it lives where you live. It can do work autonomously. So in the middle of the night while you’re sleeping go and build an entire SAS application and launch it for you. And here’s the part that separates Hermes agent from everyone else. It is self-improving. Every time you send it a message, every time you send it a prompt, it will do that task and it will learn, okay, what went well? What didn’t go well? Next time I do this, how do I do it better? And actually learns and edits its own skills. So the more you use it, the more powerful it gets. The number one question I get though is Hermes or Open Claw. Well, I got some answers for you. For those that have been watching this channel for a while, you know I was really, really big into Open Claw. I still am, but they have run into many issues that I’m hoping and I’m sure they will solve. I am good friends with the team. I know they’ll improve things. But at the moment, I am recommending Hermes Agent for several different reasons. one, it’s lighter weight. Open Claw has gotten and feels very, very bloated because it feels bloated. It’s not quite as performant. It’s not quite as quick. Hermes just feels a lot snappier and a lot quicker. But here’s the big reason. Here’s the big reason why I think you need to go with Hermes at the moment and why I think you’re watching the right video is because Open Claw’s biggest issue the last couple months is every update has broken it. Every time I do OpenClaw update, I can be like 90% sure I’m about to rip my hair out. I don’t know if they’re not testing it. I don’t know what the deal is, but every time I upgrade OpenClaw, it breaks. And for me, someone who lives with AI, who’s always using AI agents, it’s a little bit quicker for me to fix it, right? But for those who don’t use AI agents a ton, it is a pain in the woo to have to fix their agent every single time it breaks. And so that reason right there, the more reliable updates I think is the biggest reason uh to be using Hermes at the moment. They also have a bunch of other really cool things. They have the new portal which is they actually have their own AI model services built into Hermes. So if you want to save money, you can use their built-in models. It’s built to be multi- aent. So it’s really easy to spin up multiple agents at once, have them do a ton of work, have them work autonomously, which is amazing. And their social media team is just great. So, if you go on X, you have recommendations, you have feedback, you tweet it, you can be about like 90% positive that their official account’s going to reply to you within like 10 minutes. They have a really, really great social media team. Now, before we get into installation, I want to answer another question I get quite often, which is, which should you be using, Hermes Agent, Claude Code, Codeex? A lot of videos on my channel are about vibe coding, using Claude Code, using codecs. still amazing tools, still tools I use every single day. But there are times where you should be using Hermes agent and when you should be using Claude Code and Codeex. Hermes agent you want to think as like your general chief of staff, your general employee that can do any kind of work or task for you. day-to-day tasks, doing research, building documents, building spreadsheets, being your administrator for your computer, moving around files, cleaning up files, bouncing ideas, being your business advisor, right? That’s where your Hermes agent is going to be. Think of it as your general purpose business advisor, employee. It’s also great at building prototypes. Don’t get me wrong, Hermes, OpenClaw, they’re great at vibe coding. You can use them for vibe coding. I primarily use them though to build prototypes. So if I’m out on the go, I’m at the gym, I get a great idea, I pull out Telegram, I hit Hermes up, I say, “Hey, build this for me.” I come back, I have a full prototype. It’s also great at building its own tooling, right? So if I’m building like, you know, extensions to the canban board. If I’m building a mission control, I’m using Hermes agent to build the tooling for itself because it’s most familiar with how it works. If I’m doing things where I want the process to continuously improve, be better every time I do it, I like to use Hermes agent because again, the real strength of this agent overall is the self-improvement. If I just want quick changes in an app, if I want to change a button color real quick, this or that, I’ll say Hermes, go quickly do it cuz I can just spin it up real quick. You just want to think of it as like your own general purpose employee. When you use claude code and codecs, this is when you’re building largecale complex applications. Right now I’m building Henry intelligence machines about to change the world. Link down below. I’m using a combo of codeex and clawed code for that just because they have so many different features purposebuilt for vibe coding for getting complex applications built. So if you’re going deep into a vibe coding session, you have a trillion dollar idea for a business. I still recommend using cloud code or codeex for that. That’s your focus locked in work. You’re going to have open your two windows side by side codeex cla code. You’re gonna have 20 different agents going. You’re sitting down. You’re locked in. You have codeex cla code going. And end to end testing. There’s tons of really good testing features with cloud code and codec. So I still prefer again for that more complex coding session. You go with cloud code or codeex. Now that we covered some of the basics of what Hermes agent is, let’s get into actually installing it. So, we’re going to cover installing it, which models you should choose, first things you should be doing, use cases, and a whole lot more. If you learned anything at all so far, make sure to leave a like down below, subscribe, and turn on notifications. Also, make sure to join the Vibe Coding Academy. I do live boot camps covering Hermes and all the other AI tools every single Friday. Link for the Vibe Coding Academy down below. It’s the number one AI community on planet Earth. I promise it’ll be the best decision you ever make. Okay, so if you go to Hermes agent website, I’ll put the link for it down below. There is a command here, a curl command. You simply click copy on. From there, you open up your terminal. And what you’re going to do is you’re going to paste that in. Because I already have Hermes installed, I’m going to do a slightly different command, so bear with me here. Once you hit enter on the commands, you’re going to go into setup mode. Couple options here. First thing off the rip, if you already have OpenClaw installed, it’s going to give you the option to actually import all your OpenClaw stuff, all the memories, everything involved with that. Feel free to do that if you want. That’s one option. If you’re doing this clean, if you want to set up a clean agent, I actually recommend this for most people. I think using OpenClaw and Hermes side by side is a great idea. I don’t think there’s any need to have them all have exactly the same memories and skills. So, even if you have OpenClaw installed, it might be a good idea just to say no there. What we’re going to do is we’re going to do quick setup. We’re just going to configure the things we want to configure. So the first option we’re going to get is the model. Let’s discuss the different model options for Hermes. You have basically three tiers of options, right? Expensive, moderate, and cheap. Obviously, the more money you spend, the better results you’re going to get. So on the expensive side is going to be anthropic. Claude is just the best model for these types of AI agents. Unfortunately, Anthropic cut off use of OOTH for this, meaning you can’t use your existing Anthropic subscription. So, you’re going to have to do API key. With an API key, you’re basically paying for every single thing you do. This is going to be more expensive than your $200 a month for the subscription or your $200 a month for the Chai GBT subscription. So you probably if you have good use probably spend a few hundred a month. It’s going to be the most expensive but you’re going to get the best use out of it. The good news is on the moderate tier is Open AI codecs. You get a good amount of usage with your Open AI subscription inside Hermes. So even if you’re on the $20 a month subscription, you get decent usage. Couple pieces of good news there. Chat GBT inside these agents. OpenClaw and Hermes absolutely sucked up until recently. Chad GBT 54 and earlier literally completely useless in OpenClaw and Hermes. Chad GBT 5.5 though it’s very good, right? Is it as good as Anthropic? Is it as good as Clawude? No. But now it’s actually usable. Now it’s actually solid and good. So, if you’re on a tighter budget, or even if you just want to save a couple dollars a month, you can go ahead with your Chat GBT subscription here, still get pretty good usage, still get good functionality out of it, get a ton of usage. You’re going to get way more usage than Anthropic out of this, and you’ll be good to go. So, there’s nothing wrong, in my opinion, with using your Chad GPT subscription here. With Claude, it’s just going to do tasks a little bit better, and it’s going to have a little bit warmer of a personality, more fun to interact with. On the cheaper side, you have many different options. If you’re already paying for XAI, you can use your XAI OOTH here. You can also go with the new portal. This is actually the creators of Hermes Agents uh AI subscription. They kind of curate different models. It’s only like $20 a month. You’re getting cheaper Chinese models uh powering your agent, which honestly is not going to be quite as good, but it’s usable, right? Miniax, Kimmy, very usable models, very cheap models. The personality is just going to feel very robotic and it might not be that great when it comes to actually executing on tasks. I would at the very least recommend the moderate option which is Chad GBT. You’re going to get, you know, good functionality out of it. It’s going to have a decent personality. So, choose your model from there, whether it’s OOTH or API. It’ll instruct you on what to do from there. If you got to get your API key, if you need to log in, if you’re doing chat GBT, proceed with that. Now, next is going to be what messaging platform to use. And you’re going to have a few options there. You’re going to be able to choose between Telegram, WhatsApp, Discord, iMessage. Let me tell you this, the best way to communicate with your agent is without a doubt going to be Telegram. Telegram is the best way to do it. Telegram is actively working to make their platform like a platform you communicate with agents with. They’re adding tons of functionality. They’re allowing you to do a bunch of different things like topics, allowing your agents to actually talk to each other. There are so many cool things in Telegram they’re adding. They’re really the only plat messaging platform on planet Earth where they’re they’re trying to make it like AI agent friendly. You could do Discord. There’s some cool things with Discord there. I have a video on using Discord with your agents, but as a main communication channel, I don’t think anything comes close to Telegram. You need to be doing Telegram. It’s also free, so that’s pretty cool. Telegram is the way to go here. You go through those couple things, you’ll be all set. You’ll be good to go. You’ll have it all set up. When you set up Telegram, it’s going to give you a couple steps on actually connecting it to Telegram. It’s really simple. You don’t need me to put in the video. Really easy to do. You’re just copying a token from Telegram into your Hermes and you’ll be good to go. Then you’re all set up. you can start chatting with your agent. So, let’s talk about first things to do. This is your own AI employee, right? This is your own personal assistant. If you were to right now go and hire a personal assistant, they come over and it’s day one and they’re ready to do work for you. What’s the first thing you would do with that person you just hired? Well, you’d probably first tell them about yourself and your business so they’re familiar with you, right? That is what we’re going to first do with Hermes. Hermes has the best memory in the game. Hermes remembers everything about you. It’s constantly self-improving. So, let’s do this. Let’s tell Hermes all about us. So, here’s what I did, and you should do this, too, alongside with me. I am Alex Finn. I’m an entrepreneur and content creator. I have a YouTube with 200k subs and Twitter 450K. So, I told all about myself, what I’ve achieved up to this point. I built a social media tool called Creator Buddy and I’m building a new tool called Henry Intelligent Machines, which will allow anyone to build their own business. So, talking about what I’m working on and then I talk about my goals and ambitions. My goal is to make Henry Launch super successful and increase my subscribers. So, talk about your goals and ambitions as well and what you want to achieve. That’s super important because when we get Hermes to start doing things proactively for you in a second, it will be able to look at your goals and ambitions and figure out tasks to do proactively for you. So, make sure you cover those three things. Who you are, what you’re working on, and your goals and ambitions. Hit enter on that. That will be saved into your Hermes memories and it will know you so much better. And let’s do this before I start going through the user interface. So the amazing dashboard they built out, the cananband board, all of that. Let’s set up a task. Let’s set up a scheduled repeatable task Hermes can do for you every single night. So there’s this thing in Hermes called cron jobs. Basically what they are just scheduled tasks that run autonomously. We’re going to set up a scheduled task. Everything we do, everything we set up is just plain English into Hermes. It’s just plain English into the text box. You talk to it like a normal human being. It’s the best part about AI. So, what we’re going to do is we’re going to talk to like a normal human being and we’re going to tell it to do something super specific and helpful for us every single night that brings us closer to our goals. So, let me put this prompt in here. Feel free again to pause, follow alongside with me. I want you to schedule a task for yourself to do every single night at 2 a.m. It should be a micro app, UI, or system that helps us get closer to my personal goals and ambitions. It should save us time, make us more productive, or just be generally useful. Make sure this is scheduled for every single night at 2 a.m. Make sure it surprises me and makes me happy. Boom. Putting that down below. What this is going to do is make your Hermes agent proactive. It’s going to have it schedule a task for every single night where it’s going to look at memories, systems, goals, ambitions, and put something together, build something out, vibe code something for you that will get you closer to your goals and ambitions. Something that will just be generally helpful and delightful. You hit enter on that. It’s going to schedule the task. Let’s hit enter. Here we go. And GPT means this is my Hermes agent powered by GPT. I have a lot of different Hermes agents, a lot of different open claws. If you want like a video on multi- aent approach, feel free to let me know down below as well. And as you can see here, this is one of the cool parts about Hermes. You can see all its tool calls. So, it’s scheduled a cron job called create. It’s scheduled. It’s going to happen every night at 2:00 a.m. The next run is tonight at 2 a.m. And it’s going to deliver it here inside of Telegram. I set it up to create a small tangible thing every night, a micro app. It’s instructed by us towards things that save time, support your goals. Boom. Just like that, Hermes is getting to work for us. Our employee is getting to work. This is how you talk to your employees, right? You talk in deliverables. This is what I want. This is when I want it. That’s how it goes. It’s talking to us just like an employee. So, that’s amazing. Now, I have a whole bunch of other tips, tricks, use cases. We’re going to go through that in a second. I first want to take you through the Hermes dashboard. I want to take you through all the really cool UI stuff they built out. This is something OpenClaw, no one else is really doing. So, let me show you that now. And again, if you need to, you’re familiar with different things, feel free to skip down to different chapters down below. So, if you go into your terminal and you say Hermes dashboard and hit enter, it is going to open up a browser with your Hermes dashboard inside of it. This is what it looks like here. Tons of things going on. I want to kind of walk you through this real quick. If you go to models, you can very easily swap out your different models, try new models. Very, very easy to do. You don’t need to go through terminal for all of this. As you can see, my model for some of my agents is claopus, others is GPT5. If you go into your cron section, you can see some of your scheduled cron. So, you should be able to see the one you just scheduled where it’s going to do something proactively for you. If you want, you can create one manually here. This is a good way to create crown jobs. If like you want to make sure maybe it’s a little bit more complex and you want to make sure it builds it the correct way, going into the UI is a good way to do it. You also have skills. So, this is really cool. We’re going to dive more into this in a second when we go back into Telegram. But again, the strength of Hermes is the self-improving skills. Every single thing you have it do, it reviews what skills it used and how to improve the skills based on what you asked it to do. So, it’s constantly self-improving. So, you can come in here and see a lot of these skills. It has You can go in, browse, turn it on. I can even play Pokemon and Minecraft, which is unbelievable. And so, it learns things as it goes. You can see all those skills here. And make sure the right skills are turned on. We have plugins here. These are like extra powerful abilities your Hermes agent can have. Make sure you come on here and you can turn on any ones you need. It has some things you like browser use, browser base, fire crawl which allow it to crawl the internet really well. It already has uh built-in tools around using Chrome and using uh your computer computer use, but you can add on if you already have API keys around any of these, add them on here. I highly recommend enabling the image gen from OpenAI allows it to generate images. The OpenAI image generation model is the best in the game by far. So, I would turn that on so it can generate its own images as well. That makes it super good if like you’re building games with your agent. I use the OpenAI image genen model to build assets for games. So if you go to your her say, “Hey, build this game for me.” It’ll be able to go in and actually generate all the assets for the game it’s building to if you’re a nerd like me. Profiles. This is a good one if you want to set up a multi- aent approach. A profile is basically just a different Hermes agent. As you can see here, I am using something like seven or eight different Hermes agents at once. So you can go in here if you want to do multi- aent approach. Just click create, set up a second Hermes agent. It’ll be booted up and ready to go. This is cool. You can see the skills for each. So you can see which ones I’m using more. My default, which is just Hermes, has 150 skills, which means I’ve been using it more than the others. So this is a good way to set up your multi- aents config. There’s a whole bunch of options here. You can go through this if you want. Text to speech. I’m not a big text to speech guy. But where I want to take you next is canband. This is a really cool screen in here. This is probably where you’ll spend most of your time in the dashboard is the cananband board. This is super powerful and they just shipped it recently. This is basically an automated way to multitask, right? So coming in your canban board, it has the different statuses of tasks. Triage, to-do, ready, in progress, blocked, done. What’s really cool here is if you put tasks into triage, what will happen automatically is your Hermes agent will take that task, split it into subtasks, move it to to-do, and then automatically assign it to different sub aents to-do. So, if I, for instance, go in here and I say, “Script me a master class video for Hermes,” and I hit create, it’s going to go in there. And what’s going to happen in a second is the Hermes agent will actually take that, split it into steps and assign it to agents to automatically do things for me. This is really cool and this is what I highly recommend doing. I write down all my to-dos on this piece of paper every single morning. What I do is after I write down my to-dos, I go in here to this canban board and I take any tasks that an AI agent can do for me and I just put it into triage, right? I’ll just dump in like five or six tasks in triage. They go in there. I walk away. I do something else. I make breakfast. I make a pop-tart. Whatever. By the time I come back, and you saw it happen live, the tasks move to to-do. Agents start grabbing it and they start getting things done for me. By the time my breakfast is over, I already got half my to-do list done. My AI agent has already tackled half my day for me. That’s how you should be operating as well. Start every morning. kind of triage yourself with a list of tasks. You get them into your Hermes canband board and by the time you’re done your breakfast, you already got, you know, half the things you need to get done today done already. So, that’s a really cool way to use this canband board. This canband board is super powerful. You probably can find 20 other different things uh to do and how to use this. But, that’s the way I like to use just using in the morning, getting my tasks in. By the time the day is halfway over, I’m almost done everything I needed to do, which is really amazing. Now that we went through the dashboard, let’s go through a couple different use cases here. I’ll show you how it works, how it self-improves, how it uses different skills, things like that. I’m putting out over the next few days a Hermes like seven use case video. If you want to catch that when I drop that, feel free to turn on notifications down below, too. It’s going to cover in-depth beginner, medium, advanced use case. But let’s go over a few of those here. So, I’m going to show you three simple use cases any of you guys could take advantage of and get some value out of. The first one is a daily tutor. If you’re anything like me, you watch a ton of educational videos on YouTube. I’m watching them all the time. I mean, you’re watching one right now. And one thing I love to do is give those videos to my Hermes agent, have them get the transcripts, learn everything that’s in the video, and then be my tutor every morning. So, it’ll give me some fun facts from the videos I watched recently. Maybe even quiz me and ask questions. Watch how we do this. So, I got this video I put out on my second channel link down below on how to make $2 million AI in 2026. Say I’m a viewer. I watch the video. I find it super helpful. Let’s take this link. What I’m going to do is I’m going to give it to my Hermes agent. And I do this with basically any video that has a ton of content that I find helpful. I say, “Hey, check this video out. Read the transcript. Then every morning remind me of a concept or two from the video, then quiz me, and then I give the link to the video. I hit enter. And then, as you can see here, it’s using its built-in YouTube content skills. It’s going in. It’s pulling down the video. It’s reading the transcript. It’s pulling out all the important facts and information from it. And now it’s going to schedule a cron job, which again, as we talked about earlier, just a scheduled task that happens at specific times. And it’s going to send me a message every morning saying, “Hey, here’s a couple things to remind you from that video.” You should be doing this with any video or educational video you watch that has important, interesting information in it. That’s how you reinforce these facts in your brain. Got it? There’s a share journey playbook. It’s extracted 10 important concepts from it and it’s scheduled the QR for every morning at 8 a.m. and I’m going to get my first lesson at 8 a.m. tomorrow. This is how I’m learning more and more with Hermes and improving my life. Let’s get into the next use case. So, you should be setting this up. Everyone watching should be setting this one up. The next one is actually maybe the most powerful one I use, which is my computer administrator. If you put every device you got, your iPhone, your computers, your Mac minis, whatever you got, your iPad, all in the same kind of private network, your Hermes agent can move around any of them and get any content you need. Here’s what I mean by that. A lot of the times I’ll be out and about and I’ll be like, “Oh man, I wish I had that document from my computer. Oh man, I wish I had that presentation from my iPad. I can’t access it because I didn’t put it on Google Drive or anything. Now with Hermes agent, Hermes can act as the computer administrator, basically like my device administrator, and get me any file I need from any of my computers at a moment notice. This is how you do it. You go on tailscale.com. For those who don’t know, tail scale is an amazing tool. It’s completely free. It basically gives you your own private network on all your devices. Not sponsored in any way, by the way here. I just really like their product. You put it on all your devices and what’s going to happen is now your Hermes can move across all of your devices and get any documents, files, anything at all you need and move them around your devices when you need them. So, for instance, I’m on my laptop. I need an agent.md file from my Mac Studio. I just go on my Telegram and say, “Hey, Hermes, get me the agent.m MD rules file from my Mac Studio and drop it off on my MacBook Pro.” It goes, it does that. It can move around any of my devices, no matter where I am in the world, and get me the files I need. This is also great if you’re doing development, too. So, if you’re a Vibe Code or using Codeex or Claude Code, this allows you to actually test and load up local hosts from any of your devices. You don’t need to be on the device the local host is running. So that’s really helpful as well. Now Hermes is my device slashadministrator across all of my devices which is really awesome. The next use case I’ll go over is an amazing one and that is a session recall use case. What does that mean? Hermes has a feature that OpenClaw doesn’t have. No other AI tool really has which is really incredible session recall. Every session you have with your Hermes agent, every conversation you have, everything you’ve ever said to your Hermes agent is stored in a log, a session log. And that means at any point you can go and ask your Hermes to recall anything you’ve ever talked about. So if you want to say, “Hey, tell me what we talked about last Thursday at 8 a.m.” It can do that. If you want to say, “Hey, tell me what we talked about 3 months ago at 2:00 a.m.” It will be able to do that. No other AI agent has this type of memory. So what’s the use case of this? What’s what’s a good use case here? Well, we can take advantage of this incredible memory. We can say, remind me of every Hermes use case we’ve talked about a month ago. This is incredible memory. Now, at any point, you can go, “Hey, remind me that thing we talked about a month ago. Remind me that thing we talked about a week ago. What’s every YouTube video I dropped in here last week? What’s every YouTube video I dropped in here 3 months ago? What’s every link I gave you two months ago? You can quickly recall anything you’ve ever talked about, any idea you’ve idea. What’s every business use case we’ve discussed last month? You know, I am constantly dropping in business use cases to my Hermes agent going, “Oh, is this real? Can we turn this into a product?” And now I can quickly get it to recall all of those things. You can see here, look at all these use cases we’ve talked about. Uh really amazing stuff here that I can just I’ll ne I never forget anything now. everything it remembers. It’s the ultimate second brain for your life. As long as you’re running everything through your Hermes agent, you’ll literally never forget anything. So, this is a really amazing use case of Hermes agent as well, being able to remember everything you’ve ever talked about. So, let’s go into the next section, which is very particular to Hermes agent, which is the memory and skills functionality inside Hermes and what makes it so special and what makes it better than every other AI agent out there. First of all, just the basics on how those things work. All your memories and all your skills live in markdown files on your computer. And this is really important. When you have memories on chat GPT or claude, they live in the cloud. You don’t know what they are. You don’t know what’s saved. You don’t know what those AIs know about you. With Hermes, it’s different. They all live in markdown files on your computer. So everything it remembers about you, every skill it learns, you can audit and look at and change because it’s just on like a text file on your computer. But here’s the big one. Here’s what makes Hermes agent so unique is the self-improvement. Hermes built into it every step of the way. Self-improving. So whenever it does anything, every task you give it, it figures out along the way what was the best way to do that thing and updates its skills and memories for that best path. What do I mean by that? Say you tell it to go and read the transcript from a YouTube video and take out all the best practices from it. It’ll go, it’ll search for how to find transcripts in YouTube videos and then it’ll grab the transcript and it’ll find all the important points from that video. Once that task is done, it’ll figure out and review how it did that task and then save it to memory how it did it. So next time you give it that task to do, it can just go, “Oh, I did this before and this is exactly how I did it.” And it’ll do it that way again. Now, if you have them do it and then maybe it does something wrong or you say, “Hey, actually, can you do it this way?” It’ll remember that, save it to memory, and again, self-improve. So basically every time you have your Hermes agent do something, do a task, learn something, go and do something, it reviews what it did, saves it as a skill in memory, and it use that every time it does that task again. So it is constantly self-improving. The more you use Hermes agent, the better it gets, the more custom it gets to you, which is what makes it so powerful. And third, this is a big one, their memory and skills are constantly evolving. Every update in Hermes Agent adds new builtin skills. A big one from the last update was computer use. I now can have Hermes Agent go and control things on my computer. If you want to see all the tools it’s adding, all you need to do is go in, say Hermes tools, hit enter. It’s going to go in. You’ll be able to say configure and you can see a lot of the tools it’s adding here. They’re constantly adding. If you haven’t done Hermes tools yet, make sure you do that. And you can turn these on. you want to make sure you have on browser automation as well as computer use which you see at the bottom. That way if you’re ever away from your computer, you can go in and say, “Hey, Hermes, do this on my computer. Build this report. Build this slideshow.” Whatever it is, and it’ll go into it. And I know an objection I’ll get there is, “Hey, Chad GBT can already build presentations. Claude can already build presentations.” The reason why you want to be doing those things with Hermes agent is because Hermes knows everything about you. It has the best memory of any AI there. So, if you say something like, “Hey, build me this presentation,” it’ll be able to do it better than Claude and Chad GBT. One, because it’s self-improving and it learns every step of the way the way you need to do things and the way you want it to be done. And two, it knows all the context around you. So, it knows what your preferences are. So, these tools are super super valuable. Another new one they added is this video generation one. I love this one. Image generation. video generation. If you plug your Hermes into uh Gro’s OOTH or Chad GBT’s OOTH, you can now use those tools to generate assets. Let me show you using XAI Gro. I imagine generate me an AI video of a basketball player in a blue jersey wearing the number two flopping to the ground. I’m going to hit enter on that. And because it just learned this new skill, being able to generate AI videos, it’s going to be able to go in and create that video for me. So, it’s constantly learning these new skills and improving and getting better. There’s no other AI out there that can go from moving files around on your computer to getting transcripts to YouTube to generating videos. Hermes can do it all and is constantly improving, expanding its functionality. All right, looks like the video is done. Let’s see what we got here. Oh man. Yep, that number two guy in the blue jersey is certainly taking a dive there. Wow, that’s a big flopper. This is not based on any sort of a realistic person, any resemblance to anyone in the NBA, purely coincidental. So, pretty good. The next part about Hermes agent, which is really awesome, is mission control. Mission control is a custom interface for your AI agent. This allows you to build out any tool you need for your AI agent to get good work done. Everyone should have a mission control for their AI agent. One, because it’s just super helpful because you can build out custom tooling for your AI agents. Two, because it’s just fun. This is a fun thing to do. You’re allowed to have fun with AI. You’re allowed to have fun with AI agents. Not every single thing you do needs to be some sort of minmax hyper optimization to ship more code. you’re allowed to just have fun with your AI agent and the mission control is a really good way to do that. There’s a lot of custom tooling I built out here. One thing is this content pipeline where I can put in content ideas and then Hermes will go in and actually add scripts, create thumbnails, things like that. All I need to do is put a new idea in here and then Hermes takes it and moves it along the content pipeline. I have a memory page. This is basically like a memory wiki where I can go in and see all the memories my agent has. So I can go in, read everything we’ve talked about from the last month, day by day, see our long-term memory because Hermes memory is so freaking powerful. Basically everything we say and do is recorded. So sometimes I just like to almost like a journal or a diary go back and see things we worked on in the past. I had now have this really nice user interface where I can see those things. I have a docs page. This allows me to see every single artifact and document we’ve ever created. Makes it really easy to see things. Usually when you have your agent build out different documents, it goes into like a random folder on your computer. Now I can just say, “Hey, put in our docs in our mission control.” And I can quickly easily read it like this. I have a kind of an animated 2D office view where I can see all my agents walking around doing things. All my open claws, all my Hermes always working. I can come in here and see exactly what they’re doing at any given time and many more things here. All you need to do to build your mission control is literally go to your Hermes agent and say, “Hey, I want a mission control.” Build out a custom mission control where we can later add tools in that we can use to improve our workflows. You don’t need to know how to code. There’s nothing super crazy or complex here. Every time I say, “Hey, build,” someone in the audience goes, “Oh, why don’t you go into more detail?” There’s no more detail you need to know. Literally, anything you need, anything you want from your agent, just say, “Hey, do this for me.” That’s how smart the agent is. You want a new tool, go to your agent, say, “Hey, build a new tool.” Whatever you want. So, to build a mission control, literally go to your Hermes agent and say, “Hey, build out a mission control for me.” Then once you have that built, say, “Hey, okay, I want this tool now. I want this tool now. I want that tool now.” And it’ll go and build it out for you. You should have your own custom dashboard/inface just like this where you can play around, build any tools you need and have some fun. Mission control super critical concept you need to have in your Hermes agent. Next part we’ll talk about is fixing issues with Hermes agent. This hasn’t been as much of an issue with Hermes Agent as it was with OpenClaw. OpenClaw would crash every time I’d update it. doesn’t happen quite as much with Hermes agent, but you still should be prepared with how to fix things inside Hermes agent. And the way to do this is very simple. You’re basically just going to open up Hermes agent in Claude Code or Codec. So you go into Claude Code, you go into Codeex. Inside either one of those tools, you open up the folder that your Hermes agent is in. This is typically in the root of your computer. You’ll see like a Hermes folder. You open that up in claw code or codeex, then you say, “Hey, I have this problem with Hermes. Please fix it.” And your Claude Code or Codeex will go and fix that issue. That literally solves 100% of the issues I’ve ever had with Hermes. To be quite honest with you, Claude Code and Codeex are great at debugging these agents. So, any issue you have, just make sure you open up in Claude Code and Codeex and it will go and fix those issues for you. Next thing I want to talk about is security. This is a really big request for me. Security. Here’s the thing. When it comes to security with these agents with OpenClaw and Hermes, and this is going to be a really controversial thing I’m about to say. This is going to get a hundred replies in the comment section going, “Oh, you’re such an idiot. You’re so dangerous. You’re going to hurt people. Shut up.” Here’s the thing. When it comes to security with Hermes and OpenClaw, the security concerns are vastly overrated. Yeah, I said it. There I go. The security concerns with open claw and Hermes are vastly overrated. My question to you is this. If you go on the internet, if you watch YouTube, if you’re on Twitter, you probably know a lot of people that are using Hermes and OpenClaw. How many of those people have had a major catastrophic security event? Truly, do you know anybody who has had a major catastrophic security event with OpenClaw or Hermes agent? My guess is the answer you’re going to give me is no. Now, does that mean security events are it’s impossible to have with Hermes or it’s absolutely perfect when it comes to security? No, it it doesn’t mean that at all. I’m not saying it’s completely invincible. But what I’m going to tell you is this. Hermes agent and OpenClaw only do what you tell it to do. So, if you say, “Hey, Hermes, go and create me a presentation on LLMs.” It’s not going to go and then go into your private photos folder and then leak all your nudes to the internet. That’s not how it works. It’s not like you’re going to go say, “Hey, Hermes, check the score of the NBA game last night.” And then it’s going to go and leak all your text messages on Twitter. That’s not how any of this works. These agents only do exactly what you tell it to do. So, the only way it’s going to have a major catastrophic security event is if you go and you tell it to do something extremely stupid, if you go and tell it to delete all your text message and all your emails, it’s going to delete all your text message and all your emails, but nothing else. If you tell it to go and drop you a document in your folder, it’s going to drop you a document in the folder. It’s not going to do anything else other than that one thing. So, what that means is you need to have some accountability, right? You need to think before you give it a prompt, is this going to cause it to do something really stupid? Am I asking it to do something really stupid? It’s not like you can say, “Hey, code me an app.” And then it’s going to text your ex-girlfriend. That’s not how this works. So, just think before you go, is this thing I’m asking you to do bad in nature? If not, then you’re good to go. You have nothing to worry about. But for some reason, these naysayers and doomers on the internet say, “Oh, Hermes, it’s the most dangerous thing ever. Don’t put it on your computer. Don’t put it on your main computer. No, put it on your main computer. I have it on my main computer. I have it on every computer I own. I didn’t isolate it. I don’t think you need to isolate it. I think going and creating a new iCloud account just for your Hermes agent is actually a quite stupid and timewasteful thing to do. I don’t think there’s any point in doing it. As long as you just have some personal accountability and think about what you’re telling your agent to do, I don’t think there’s any risk at all. And I know that’s going to piss people off. I know the purists online who’ve been going around telling everyone, “Oh, this is a major security threat.” That’s going to make them go to the replies and cry like little babies, but I don’t care. Until someone on planet Earth can tell me someone who’s had a bad security event with one of these agents, I don’t really care. Just use good judgment. Be an adult. Think about what you’re doing and security won’t be a concern. I don’t think you need to make an iCloud account. I don’t think you need to make a separate Google account. I think you’re adding radical amounts of complexity for no reason. I think you should install it on your main computer. I don’t think you need to have a VPS, which for the record, the VPS’s are scams. Don’t put on VPS’s. Every AI YouTuber on planet Earth is getting paid by Hostinger to shill you a VPS. I promise you all those AI YouTubers who are shilling hosting jer literally zero of them are actually using hosting jurer themselves. So don’t listen to them. Just install Hermes on your main computer. I don’t care. You’re not going to have security issues if you practice some accountability and good judgment. That’s security. You don’t need a Mac Mini for this. You don’t even need to buy like a super expensive Mac Studio like I did. All you need is whatever computer you own. Could be an old dusty laptop in your closet. be whatever you’re using right now. You can install it on there. There’s no concern. There’s no issues. I all of a sudden expose all your text messages. Relax. That is the basics of Hermes agent. If you watch this video, you know how to set it up, what to do, how to run it, what you should be doing with it. I make tons and tons of content on AI agents like Hermes, like OpenClaw, like Claude Code, like Codeex. So, if you learned anything at all, like, subscribe, turn on notifications. I do live live live boot camps every single Friday in the number one AI community on planet Earth, the Vibe Coding Academy. Link down below for that. Make sure to join. We have thousands of people in there that absolutely love it, that are learning a ton. If you want to stay on the cutting edge, join it. Best decision you’ll ever make. I am so deeply and eternally grateful that you would spend all this time with me to watch these videos and learn from me. It’s an honor of a lifetime. I can somehow build this platform and make you educational AI videos. I’m so grateful.

I’ve built hundreds of Poly Market trading bots. But the easiest bot to build is actually [music] not a prediction bot. It’s a copy trading bot. But the version I want to teach you how to build here today does not just find a whale and blindly follow whatever they do [music] because that breaks too fast. A wallet can show massive profit and still be a terrible copy target. Maybe all the profit came from one lucky trade or a few lucky trades. Maybe the wallet is only good in one category. Maybe the markets are too illquid. Maybe by the time you [music] see the trade, the price has already moved way too far. So this bot starts one step earlier because before it copies trades, it finds [music] the wallets worth copying. It pulls the Poly Market leaderboard, scans the top 500 wallets, studies the last 30 days, scores each buy ROI, consistency, and copyability. Then it trades the best signals [music] with simulated $5 to $20 positions. The whole thing runs through Hermes agent and the performance shows up in a dashboard inside [music] of my own mission control. Hermes scans the leaderboard, updates the wallet profiles, watches new trades, and basically runs [music] almost everything for us, and then sends us an end of day report. Look, I’ve had strategies make me thousands a month, and I’ve had strategies lose me thousands [music] a month. By the end of this video, you should have yourself a strategy that improves with time on its own without you having to do anything but monitor it every once in a while. If you keep watching this video till the end, I’ll also be giving you the full prompt to build out everything I’m about [music] to outline in this video. So, this is not just a trading bot. It’s a copy trading research [music] system that has to prove the edge before it earns autonomy. And the first problem is not how to place trades. The first problem is figuring out who is actually worth copying.

Let’s get started. The beginner mistake is thinking the leaderboard tells you who to copy because it doesn’t. The leaderboard tells you who made the most money. And that is very useful information, but that’s not enough. A wallet can make money in a way that is impossible for you to follow. If one wallet made most of its profit from one massive trade or from like three big trades, that is not a repeatable signal yet. If one wallet trades tiny markets with no liquidity, the return might look amazing for that wallet, but copying it could be impossible. And if one wallet enters early and moves the price too high, you might always arrive late. Something very important. If one wallet is great at, say, politics, but terrible at crypto, copying every trade from that wallet, including the trades or the categories it sucks at, is just a recipe for disaster. So, the first part of the bot that I recommend you build out is a leaderboard scanner that pulls the top 500 wallets from Poly Market. Then, it does not ask who made the most money. It asks who is copyable. Here’s the difference. Let me explain. For every wallet, the bot should calculate three main scores. ROI, consistency, and copyability. ROI tells us whether the wallet made money. Consistency tells us whether the wallet seems repeatable or whether the results came from just a few lucky hits. Copyability tells us whether we can realistically follow the wallet without getting wrecked by uh late entries, bad spreads, or just category mismatch. That score is the real starting point for any of your bots because if the wallet universe is bad, the rest of the system is just beautifully organized chaos. The bot should also rank wallets by category. Politics, crypto, sports, macro, so on and so forth. A wallet should not get treated as universally smart just because it was right in one lane. That is where a lot of copy trading systems get sloppy. They treat a wallet like a genius when the wallet might only have edge in one specific market type. So before the bot watches trades, it creates wallet profiles and whether blind copying that wallet would have worked over the last 30 days because you always want to have comparison, a baseline that gives the bot a better foundation. But once we know which wallets are worth copying, we still should not blindly copy every trade. That is the next mistake. The dumb version of copy trading is simple. Wallet bought yes. So I buy yes. That’s not a strategy. That is outsourcing your bad decisions to someone else’s wallets. A better copy trading bot starts with a better question. Is this trade still worth copying right now? That means every new wallet trade gets scored before anything happens. Who made the trade? What category is this wallet actually good for? What price did they get in? What price can we get in now? How far has the market moved since their entry? How soon does this market resolve? And is this wallet usually early enough to copy or is this a real thesis, not just signal? The decision labels that your wallet uses should be simple. Copy, watch list, or skip. Copy means the trade is strong, the category is a fit, the market is liquid enough, the entry is not too late, it passes all the checks. So, we copy. Watch list means the signal is interesting, but the entry is not clean yet. So, the bot watches it. It does not copy it, but it places us on its watch list to see how it’s done later on. Should we have copied it? Was it a good decision not to copy it? So on and so forth. Skip. The last one means the trade is too late, too liquid, too noisy, or just outside that wallet’s proven edge. This is where the bot becomes super useful. Not because it predicts everything, because it refuses most of the bad copies. A good skip is as important as a good entry. Trust me when I say this. If the bot stops you from chasing a wallet after the market already moved like 25%, that is valuable. If it ignores a high ROI wallet because the wallet only made money from a few small trades, that is valuable. If it says a wallet is interesting but only in crypto markets, that is valuable. But filtering one trade is still not enough. The real edge comes when the bot tracks whether its filters actually worked. That is why the next layer is paper trading. Okay, hear me out when I say this. Ask someone who’s done this hundreds of times before. The first version of your copy trading bot should not place real trades. As much as you want it out in the wild making you as much money as early as possible, it should paper trade first. Paper trading means the bot acts as it is placing real money, but it does not actually execute anything. It just simulates that it is. If a high-scoring wallet enters a market and the trade passes the filter, the bot creates a simulated position, it assumes as though it’s placed money on it and then it tracks. By the time of closing, it sees did we make money, did we lose money, was it a good bet, a bad bet. The way I typically place my copy trading wallet is the following. I have three tiers. It places $5 bets for a decent signal, $10 bets for a strong signal, and $20 bets for the highest confidence signals. This mitigates the risk so that you’re not just $20, $20, $20, and you end up wrecked after a couple of uh not days, hours. Then Hermes updates the paper P&L every single hour. The bot should not only ask, did we make money on paper, it should ask, did our filtering improve the strategy. If blind copying the leaderboard beats our bot, then the bot is not adding value. If the filtered strategy beats blind copying, now we have something worth studying. I also want the bot to track missed winners and avoided losers because those are both lessons that it can use to learn and improve. If the bot skips a trade because liquidity was bad and that trade later loses, that was a good skip. If the bot skips a trade because the entry looked late or bad and the market keeps running for some reason, maybe the late entry rule we have is a bit too strict. The point is that paper trading creates evidence and without evidence, the bot is just confident for no reason. With evidence, however, the bot can improve and the evidence lives inside the decision journal. This is the soul of your system. Every signal needs a decision journal entry. Not just what happened, why the bot made the decision. For example, wallet A bought yes on a crypto market at 42 cents. The current price is 45 cents. The spread is 3 cents. The wallet has a strong crypto track record. Market resolves in two hours. Liquidity is acceptable. Decision copy. That is what a journal entry should look like. Uh along with a reason. This wallet was strong. Uh good category fit still early enough. Price movement is not too far. Your bot should place its reasoning for copying the trade. That journal is what lets your bot get better. Later, when the market moves or resolves, the bot can review all of its previous decisions. Did paper copy trades actually improve after the signal? Did watchless traders become better entries later? Did skip save us from bad trades? This is where the self-improvement part of this video actually happens. Maybe the bot learns not to paper copy if the price has already moved more than 12 cents. Maybe it learns that a certain wallet is only copyable for crypto-reated events. Maybe it raises the minimum liquidity threshold. This is what self-improving should mean. Not the agent magically becomes a genius overnight decision loop that tracks its own reasoning and gets less stupid over time or smarter over time, however you want to see it. And because this is paper trading, I recommend the rule updates to happen automatically. Again, the bot is not trading real money at this stage yet. So, it should not ask you before changing a paper trading threshold. It should just log what changed. Then why rule version one becomes rule version two. Uh the dashboard shows the old rule, the new rule, and the evidence and the reason Hermes changed it. That is the difference between a notification bot and an agent. A notification bot tells you what happened. An agent changes how it decides based on what happened. Now the question is what we actually need to build. On to step five. I recommend you build this in nine parts. And again as I promised at the end of this video I’m going to give you the full prompt so that you can build this out yourself. But when I say I recommend you build this in nine parts, it’s me saying I recommend that you check each of those nine parts work step by step chronologically. So don’t fix step five before step one works. That being said, let’s start with one. First thing is the leaderboard scanner. It should pull the top 500 wallets and create a wallet universe. It’s just, you know, a universe where wallets exist and are being tracked. Second is the wallet profiler. It scores ROI, consistency, copyability. It takes every single thing you would want to take a look at and it gives it a score so that you minimize copying traders that are one hit wonders or that make just a few good trades or that impossible to copy because they trade at the very last second of a market. Third is a trade monitor. This should watch selected wallets and detect new trades. The fourth would be the trade scorer. It should check whether the new trade should be paper copied, watched or skipped. Right? Copy, watch, skip. Fifth would be the decision journal. This is again where it records the score, the reason, the risks and the rules that it changed every time it decides to adapt. Sixth is the paper trading engine. It should create simulated positions. That’s up to you whether you want it to be like mine where it trades between5 to $20 positions or if you want to just yolo and go straight to $100. It’s paper trading anyways. So really, you can choose whatever you want. I recommend you simulate it as though it is real. And say you want to place $100 inside your wallet once it goes actually live and starts trading actual money. Then you know don’t simulate $100 per trade. Simulate as though the wallet has $100 in total. So simulate reality. Seventh would be the hourly P&L updater, profit and loss updater. Uh you should have a system that checks prices and updates your P&L every single hour. Eighth would be the outcome reviewer. This should check what happened after 1 hour, after 6 hours, after 24 hours and the final resolution. And [snorts] ninth, finally, the rule updator. It reviews performance and automatically updates scoring rules with version history, aka how your bot improves with time. That is the actual system, and Hermes is the operating layer sitting on top of all of that. But if Hermes is running the loop, chat is not enough to supervise it. That is why we move on to one of the most important pieces of the puzzle, the dashboard. I want my copy trader to live inside my Hermes agents mission control. I call mine Hermy HQ. Uh Telegram is good for summaries, don’t get me wrong. Hermes is good for doing the work, but performance needs visualization. We’re visual creatures. It her my Hermes could send me a block of text and I could be like, “Okay, cool.” But just seeing that P&L go up is what signals to my ape brain that, “Oh my god, this is working.” The dashboard should answer three questions immediately. Are we profitable on paper? Which wallets are worth copying? And what did the bot learn today? That’s it. You shouldn’t get analysis paralysis by looking at thousands of different things on your screen. There shouldn’t be a giant wall of text or charts that makes you feel smart but does not change the decision. The overview page should show paper win rate, open paper positions, active tracked wallets, copy candidates today, latest rule changes, and an end of day report status. It’s your dashboard. So, if there’s something that you prefer to see or something that you prefer not to see, then feel free to change that. This is where you can get creative. This is where I recommend you get creative because the dashboard really should be something you enjoy looking at. You enjoy watching, you enjoy opening up and seeing whether you’ve made money or lost. For the actual build, I literally asked Hermes to refine my initial copy trading agent and this is what it gave me. Build me a Hermes powered self-improving polyarket copy trading bot with a Verscell dashboard that can be added to Max HQ. This is my uh mission control, but for you, I would change it added to my mission control. Or if you don’t have a mission control, here’s where I recommend you just remove that with a dashboard. Versel doesn’t matter. So feel free to customize this depending on your situation. Uh and then as you can see, this is a very very very [clears throat] long prompt. it will take. Hence why at the beginning of the video I said an hour of back and forth because there was a lot to build and focus on. That being said, uh what I did, I took this and I copied it onto my Hermes agent so I could build one out for you guys. Literally here you can see me pasting that massive mammoth of a prompt and then it literally running through and building it out. And this is what it looks like now that it’s done. It started trading at 9:00 a.m. today. It’s currently 11. So, it’s been two hours exactly that it’s been live. And as you can see, it just started and it’s updating its positions accordingly. Uh it’s currently $5 in profit. But again, uh this is paper trading right now. This is where it starts developing its uh thesis. it starts improving every single day, every single week. And at this point, you just want to let it run. Let it improve. Let it find a profitable strategy or profitable strategies plural. And then you can start running it. This is essentially what it can look like once it is completely uh live, once it’s learned a lot more, once it’s developed its own strategies uh essentially. And as you can see, it can start improving so on and so forth and scoring itself after every uh improvement. And as you can see, not only should it show you your P&L, it should show you your win rate. Should show you which version it’s on and what it has learned with every single version, with every single new version that it implements. And that’s pretty much it. That’s what it should look like. By the way, quick update here. Just before I go to sleep, uh I wanted to quickly check on our copy trading bot. And this is where we are at. We are $27 in profit with a 57% win rate. It’s only copied six positions today. Yesterday, uh just exactly when we set it up, it was at 7 uh just right before we sleep. So, it’s been 24 hours. And as you can see, right, started $5 negative, then went up to eight, went back to break even, and boom, all of a sudden almost $30 in profit. I went ahead and asked my Hermes agent what was up, uh, what was driving that, and it’s mostly some of these trades, which are USA versus Australia. Uh the match just happened today and our bet won which won us $15. Again, this is paper trading money, so we don’t actually make anything yet, but just goes to show the strategy is on uh a good turn. Our bot is currently self updating and on version three. So yeah, this is just version three. The long-term goal for you should be autonomy for the bot to be able to run on its own, but autonomy has to be earned. I would not let this place real trades just because the dashboard looks cool. I would want at least 30 days of paper trading, positive paper trading, at least 100 paper trades, a clear win uh over blind leaderboard copying. no major data failures and a draw down profile that does not look insane. Until then, this is a research agent and that is not a limitation by the way. That is the point. If the paper version proves the edge, then real execution becomes a separate problem. [snorts] If the paper version fails, then the bot saved us from automating a losing strategy. And that my friend is also a win as well because the goal is not to make an agent gamble. The goal is to build an evidence layer before money is involved. Look, a good polyarket copy trading bot is not a whale following bot. It’s a filtering system. That is the part most bots skip. They chase signals, but they do not review their own decisions. And if you do not review decisions, your bot simply can’t improve. So the system you should follow for any copy trading bot is very simple. A leaderboard scanner, a wallet profiler, a trade scorer, a decision journal for self-improvement, a paper trading engine for testing validation, a rule updater, and a dashboard for ease of access. If you want to build this yourself, I will put the full build prompt in the description. But please, please, please make yourself a favor. Do yourself a favor. Start with paper trading first. Make the bot prove the edge. Then decide whether it deserves real

Good morning. Uh, welcome back. We’re at Best Buy today. Christmas came early. More like I had a chore to do. I needed to buy a thing for maintenance reasons. It just so happens to be around Christmas. Nonetheless, I’m going to run into Best Buy right now, grab something, and we’ll talk about it later. All right. All right. We are back from Best Buy. Let me show you what I got.

20 24 terb of storage. Not my first 24 tab of storage. This is actually only for backup. Um, let me explain what I plan to do with this. So, yesterday I’m driving into work. I have a whole plan for this video that I want to shoot. And uh Jeremy the OG hits me up and goes, “Hey, how do you get your Mac Mini to uh be your file server remotely? All these things.” He knows that’s what I do for file management. And he said, “Is there a YouTube video that I could go to and look at to set this up myself?” They say, “Whoa, whoa, whoa, whoa. YouTube video to learn about this thing I’m already doing here. I am a YouTuber. I’ll just make a video for you. He then called me a content cow. You know, whatever. I’m making Jeremy this video as an instructional video for how to set up a Mac Mini as a remote server for file storage. basically replacing what a NAS would do. So, so back to Best Buy. So, I already have 24 terabytes in a RAID 5 system serving me all my media, but I’m paranoid and I need some reinsur reassurance. And so, I bought 24 TB for backup storage. I would say most notable, most intelligent people on this on the internet would tell you you want to have three copies of your files in two locations and one in the cloud. I think that’s the 321, something like that. You know, I’m not made of money. I can’t do that. So, I’m progressively getting there and hopefully nothing goes wrong. Fingers crossed. Um, until I get to a proper 321. Today, I’m achieving part of that. So, I want to set this up as a backup time machine setup onto my pre-existing system. First, let me show you how I have everything configured. I’ll do a little instructional for Jeremy. Shout out the boy. So, pretty pretty simple setup. We have the Mac Mini plugged directly into my network with a Thunderbolt cable coming down here to the OWC Thunderbay, uh, which lets me do RAID 5 and allows me to do five HDD hard drives. Um, it’s really noisy. I bet you can hear it right now. That is my one con my one complaint here. Uh but then everything is managed via Mac OS software and OWC software. All right, let me show you how I have this set up. As you can see, I have Mac Mini as a server option. Um and it it shows me all my Macintosh HD stuff. It shows me the OWC Thunderbay, which I call SAM server, and then a user folder. How do we set this up? How do we get here? All right, here we are. This is Mac OS 26. I set this up on 25. It should be the same deal. No, it wasn’t 25. What What did they call it before? Not Tahoe. Sequoia. I did this on Sequoia. Go into your settings. general sharing. And we want to turn on file sharing. Boom. Okay. I’m going to blur this IP address, but there’s an IP address, an SMB IP address. That is a super important number that we’re going to use. We’re going to add shared folders. This is where you can pick your external hard drive. So in this instance, we’ll choose Mac Mini SAM server. Then what we’re going to do, we’re going to turn off allow full disk access for all users. And we should be able to add users. Now

we’ll we’ll There we go. Okay. Sam Brown read and write everyone else no access. So using my computer password login I will have access to this uh server. That’s the core of it. That is as simple as it gets. You just turn on file sharing in general. You make note of that SMB server uh that SMB SMB IP address and then you go into your finder and you connect to a server and bam, you can just type it in here. Here’s the big fork in the road, Jeremy. This is for you. This only allows you to access files and your server in your network. So, you have to be on Wi-Fi in this place for it to work. And you’re like, well, that’s not as helpful. The goal for me when I was setting this up was to be very similar to iCloud Drive. I love iCloud Drive. I think it’s up there. It’s It’s so nice. the integration, how it does downloads and uploads and stuff. It’s near perfect, but I’m capping I think I’m on the family plan with like six terabytes in total and here I am with 8 tab trying to fill 24. It just I can’t I’ve outgrown it. I need something else. The biggest aspect of that is remotely accessing your files. How do I do that? Pretty easy. I go to a website called Tailcale. Tail Scale is great. It’s a free-touse VPN service meant for businesses and most importantly this act this application of file sharing. Here you’ll see I have four machines all connected to tail scale. Um, and then I set my Mac Mini as an exit node. So, because I use the exit node on the Mac Mini, it emulates my houses network anywhere I’m connected to my tail scale VPN. that then means I have to use tail scales IP address. So when I go to connect to a server I use uh this like 100 blank blank blank not the 192 that is in my house. I don’t want to go too much into tail scale cuz honestly I don’t know a ton about it. I know enough to get my VPN set up for this file sharing and stuff like that. Um, not enough to recommend for other use cases apart from this. But everything I’ve done with Tail Scale has been kind of awesome. It’s free to use. I have four machines, my iPad, my phone, my MacBook, and the Mac Mini all connected. And the Mac Mini is the hub and it sets my my IP address for when I’m remote and have to still access my files and stuff. So, I’m loving it, but I I don’t want to go so far as to recommend it. Any VPN will work in this situation. Just requires a little more like uh

a little more study. Um but yeah look into tail scale Jeremy use tail scale. There is a third aspect to all of this that is the jiniqua. Uh the Mac Mini is very good at staying on for long periods of time but you know you have power outage things happen. That’s why it’s connected to the TV and I can access it remotely or go into the actual computer and use it like a regular computer. But to remotely access I use Parseek. Parseek also a freeto use um program. Again, sorry I’m blurring everything. I just don’t want there’s work stuff. I don’t want bad actor, you know, to see any of this. But I can connect to my Mac Mini, which is running soft RAID, which puts me in RAID 5. And you can see I have 24 terabytes. They’re all kicking. That’s it’s as easy as that. I mean, you know, your specific use case, you know, you’ll find there there are things that you got to adjust and whatnot, but overall,

it’s as easy as that. This is a DIY solution compared to Synology, UG Green, um, who else is making Naz’s dedicated Naz.

I’ve heard kind of bad things about them. I know my system is not like the most ideal setup, but it works pretty amazingly for what I got going on. So, yeah. Uh Jeremy, let me know. If you have questions, hit me up. If you are watching this and are thinking, “This is cool. I have a Mac lying around and I want to, you know, host my own data without using Google Drive or iCloud or buying Sonology and do this. Let me know how it goes. Uh it’s been really fun project to work on and it’s not over. So that brings me to today and what I’m doing with this expansion port. I don’t have 24 terabytes lying around to serve as a backup. So, I bought this thing. It was on sale at Best Buy. I want to set it up with Time Machine to be a

automated backup system that I never have to think about. It doesn’t need to be RAID. It just needs to be another chain of foolproof. So, let’s try and set that up now. All right, we just got this guy set up. Um, super simple. It just connects via USB 3.0. I wasn’t expecting that, but you know, it’s it’s HDD, so there’s not like a ton of throughput here. Um, let’s go back to the computer and see. Let’s get this guy set up. All right, we are back into the Mac Mini. Let’s see.

See what we got.

Okay, just like that. Expansion, it’s there. Let’s get this guy initialized. Let’s go to disk utility. All right, you can see these are all my RAID drives. Seagate expansion, though. That’s what we are going to mess around with. We’re going to erase that. All right. So, I’m just erasing uh the new expansion drive before we set it up in Time Machine. All right. So, it just erased and you can see now we have Sam’s server back. What happened there? SAM server backup.

Perfect. Perfect. And just to check we didn’t erase the wrong thing. Sam server is here. It’s got everything I need. All right. We’re going to add a time machine backup now. SAM server backup. Setup disk.

Encrypt backup.

Uh, no password. That’s okay. Disk usage limit none. Done.

All right. Uh, was easy as that? I I’m not sure. Uh Sam server Sam server backup is there. Open time machine settings.

Oh, you know what? Let’s exclude. I see this is what I want to exclude. I want to exclude Macintosh HD.

Bang. There we go. Excluding Macintosh HD, we have 6 78. It’s climbing, but look, only.1% done. It’s got to back up 8 GB of storage right now. Um, but once that’s that initial aspect is done, it’s going to do it every time there’s a change. So, it’s not I don’t have to manually back this up every time I do something. It’s going to have everything copied over and once something changes within the volume, it’s going to just write that little section. So, should be super simple. Should be super easy. Only have to knock out this first 17 hours and everything else is a breeze. But yeah, that’s it. That is my file management setup. I use tail scale. I use a Mac Mini and create my own NAS um using a DAZ which is direct attached storage with the OWC Thunderbolt 4. This is all just summary. Hopefully you caught it. Hopefully this is helpful and you get to do something similar, you know, repurpose an old Mac that’s lying

大家好，我是尼卡拉塞昂。最近我深入研究了一下五角大楼的449号大案，又被震惊了。这个大案里讲了1961年FBI接到举报，说有一种自称为Corindaurian的外星人，正在通过无线电联系人类。我看完这份FBI的报告，全身都起了一层鸡皮疙瘩。这个大案中记载的Corindaurian外星人，他们的科技能力还有对宇宙的理解，根本就是释迦牟尼在《长阿含经》《法华经》里面讲的佛国的翻版，一模一样。很有可能他们就是佛经中提到的天人异族。这种科幻加外星人、玄学加佛学的理论在五角大楼的这个大案里出现，真的是让人相当惊讶。

今天我就给大家讲一下这故事。事情发生在1960年代，当时整个世界处于冷战最紧张的年代。有一天，FBI接到举报，说有一个18岁的少年，自己在研究无线电和不明人士通话，里面的语言充满了反对越战、反对美国政府的信息，所以爆料人员怀疑这是苏联级别的渗透。FBI一听不得了，赶紧立案调查。于是这些探员们就开始调查，但这一查他们全都懵了，因为这件事太玄幻、太科幻了。因为这个18岁的年轻人说，和他通话的无线电不明人士是来自400光年以外的外星人，叫做Corindaurian，他们的领袖叫做Master Callanlee Rutan。这个Master大师不但给他解释了宇宙文明，还安排他去了母舰聊了很多外星科技，其中还提到宇宙宗教等等。这些都是些不着边际的话，让FBI探员非常无语。

被调查的年轻人叫做Bob Reynolds，他从小就对无线电非常感兴趣。在60年代，全世界、美国都没有什么手机，没有网络，美国年轻人玩的都是业余无线电。这在当时相当于今天最厉害的极客高手。1961年7月初，Bob在自己的房间里用自己组装的无线电探索全世界的节目时，突然扬声器里就传来了嗡嗡的声音。一开始Bob觉得这些杂音是他自己DIY电台里出问题了，于是就开始检查、换频段之类。但是每次换都有这个嗡嗡声，而且越来越响。这个时候他反其道而行，不躲了，反而把频率精准地一点点往这个噪音源对上。接下来的震惊就来了：在他把频率严丝合缝地对准噪音的一瞬间，嗡嗡声、哔哔声突然就停了，无线电里突然传出一句话：“Bob，我们希望你能在这个频率上多停留一会儿。”

Bob说这个声音是一种柔和、温暖、像水晶一样清澈的女生，说的是英语，而且在无线电里直接说出了他的名字。Bob愣住了，大脑一片空白。接着女生就说：“我是Lin-Ely，我和我的伙伴来自一个叫做Corindaur的行星，我们现在就在你们地球上空几万英里的飞船上，正在和你通话。”在档案中记载，Bob说这个女人说话的声音就像唱歌一样，有一种美丽的音乐感，非常有节奏。她的描述非常诡异，因为《华严经》里面就讲“梵音微妙，令人乐闻”，梵天的声音都是非常微妙的，听了让人非常快乐。在《佛说大乘菩萨藏正法经》中记载“迦陵频伽微妙音声”，这迦陵频伽是什么？它就是佛经中的天人，经文中记载他们生活在西方极乐净土，声音清婉，因天人紧那罗一切鸟类所不及，这鸟叫的声音都像音乐一样，非常有节奏。

我之前讲过梵语的秘密，还有海奥华人的秘密，里面都提到过，除了那些什么小灰人之外，对人类友好的外星人，说话都像唱歌一样有节奏，柔和，有一种咒语般的音韵美。那个外星人Lin-Ely就解释说：“我们算准了像你这么对科学有好奇心的年轻人，一定会去寻找杂音的源头。”所以Bob被震惊了，说这给我当头一棒，重重的一击，我惊呆了，这种震惊持续了整整一分钟我才回过神来。

接着这个外星人Lin就开始向Bob解释他们为什么来地球。她说：“我们不是来侵略地球的，我们一直在观察你们。我们联络你是因为我们有强大的读心和远程心理探测技术，在对你观察一段时间后，发现你对其他星球的生命有着强烈的兴趣，内心深处渴望世界和平以及人类的美好未来。我们需要寻找一个地球上的平凡代言人来传递宇宙中的和平和博爱。我们迫切希望看到人类在和平中生存下去，而不是像一个正在玩火快把自己烧死的儿童。”

从Lin的话来讲，还有之前我讲过的外星人档案，我们会发现为什么外国关于外星人的记载从1950年开始变得特别多，而且除了小灰人绑架人做实验的故事外，和人类友好的外星人都在传递让人类放弃战争、渴望和平的信息。这个是有历史大背景的：从1945年美国成功试验核弹，8月投了两颗原子弹在广岛和长崎，之后从1947年飞碟目击案件就开始增多。到了六七十年代，就是Bob生活的年代，美苏进入冷战高峰期。当时核军备竞赛已经到达只要稍微差一点错，整个地球文明都会消失的阶段。当时美国和苏联天天疯狂测试核弹，有在海里爆炸的、天空爆炸的，还有核弹超越大气层在近太空核爆的。当时美国最著名的一次核试验叫做“海星一号”（Starfish Prime）。1962年7月9日，美国从太平洋用雷神火箭把一颗当量大约140万吨TNT的核弹送到离地400公里的近太空点燃核爆。爆炸瞬间产生强力电磁脉冲，夏威夷的电力和电子设备全部出现电磁故障，而且天空直接炸出了人造极光和人造辐射带。

在这个之前，1958年美国还搞过一个叫做“阿尔戈斯行动”，专门测试核爆能不能在地球磁场里造出人工辐射带，目的就是为了干扰、摧毁卫星和地球通讯。所以苏联一看美国这么干，它也没闲着，开始了一个核实验计划叫做“K计划”。从1961年到1962年，前苏联从卡普斯京亚尔发射了无数导弹飞向哈萨克斯坦方向，并在高空引爆。最吓人的一次就是在1962年10月22日，就是古巴导弹危机最紧张的那几天。这次核实验在高空核爆产生的高强度电磁脉冲，直接就破坏了哈萨克斯坦境内所有的电力线和通讯设施。所以就是在1962年前后，美苏接连的核爆，整个地球的海陆空甚至太空都放出了核爆烟花，外星人都看不下去了。从1947年开始到70年代，超多外星人联络地球的档案记载都是劝人类不要再玩火了，要不然地球真的完了。

在很多人类接收到这个信息的时候，这个事情也非常巧合：1943年之后，美英苏就签署了部分禁止核试验的条约，把大气层、外太空和水下的核试验都给禁了。1962年之后，到了1963年还出了一个大事，就是美国总统肯尼迪被刺身亡。在63年的档案中记载，Lin和Bob接触之后就说：“你们的参议员除了少数几个，都是照本宣科的机器人。你们的总统不幸只是他们需求的一枚棋子。肯尼迪先生所遭遇的命运，因为他出于伟大的人道主义，不愿为这军工产业这‘八爪章鱼’牺牲自己的理想，最后他为人类付出了最高的代价。”不知道她为什么提这个八爪章鱼，这八爪章鱼就是漫威里面的九头蛇，它的标志就是一个骷髅和八爪鱼，所以外星人就是暗示肯尼迪被杀是美国体系内的深层政府的神秘人干掉的。

以上这些信息一开始都是Bob和外星人Lin用无线电进行通讯。在一两次后，Lin就开始教Bob把无线电和电视连接起来，魔改成可视电话。1962年1月6日凌晨2点，Bob戴着耳机盯着他改装好的电视屏幕准备测试。当他打开电视的一瞬间，一个环形的房间形象显示出来，中间坐着一个绝美的金发女郎，蓝色眼睛，看起来大约18到20岁左右，穿着好像滑雪服那种柔软的两层套制服，头发上还系着两条蓝色丝带。这个金发的“外星人”就是我们以前讲过的Hyperborea（海波利安）北欧型金发外星人。佛经中的北俱卢洲人、上古人类和外星人混血的雅利安人，都是他们的后裔。雅利安人是不吃肉的，现在印度的高种姓信的印度教，当年婆罗门大祭司们的后裔，他们也保存到古代的习俗，都是不吃肉的，每天就是吃咖喱蔬菜和碳水。

Lin介绍说他们这个种族Coreandaurian人是整个宇宙行星联盟的最高行为准则。他们对宇宙所有的生命都有着爱，在他们的伦理观里，任何形式的暴力、剥削、杀戮都是极其原始的，违背宇宙法则。所以他们也不吃肉，Lin说他们吃的是一种非常完美的合成食物，可以通过原子复制机直接打印出完美的营养合成物。Lin说他们的星球叫做“克雷多星”，在牧夫座的方向，距离地球大约400光年，有一个恒星叫做“克雷娜”。他们的行星有四个月亮，他们说的语言叫做“加林瓜语”，而这种语言是人类最原始的语言，地球上所有的语言都来自他们的加林瓜语。Lin说在人类《圣经》中创世纪记载的巴别塔都是真的，当时的人类用非常先进的技术创造了一座巨型高塔，但是因为它太重了，最后引发地质塌陷倒塌。而且也说这个之前的“大洪水”也是存在的，是彗星引发的气象灾难。

然后Lin又解释说：“你们的人类文明是地球历史上第三代了。之前地球上曾经出过两个极其辉煌的文明，就是你们传说中的亚特兰蒂斯和利莫里亚文明。根据他们的档案记录，就是在最后的恐怖战争里把自己彻底炸得灰飞烟灭。”他暗示这两次文明都开发出了核武器，然后进行了核大战，也许他说的就是《摩诃婆罗多》里面记载的各种事情。

接着Lin跟Bob爆了一个大料。她说：“你们人类崇拜了2000多年的耶稣基督，其实是宇宙行星联盟派来的一位高度进化的星际大师。他来地球只有一个目的，就是向野蛮的地球人传达宇宙的兄弟情义和和平的信息。”这个我以前也讲过，在UFO圈流传的一个计划，就是美国政府1950年进行的一种叫做“水瓶座计划”里，就提到过说处女玛利亚怀孕、耶稣、天使报喜等等，都不是神话，是一场被称为“基督生成计划”的星际基因和道德挽救实验。耶稣也不是地球人，他是统治某宿星团的第12位最高长老之一，他的星际间真名叫做“本·约瑟夫·约舒亚”。他们在地球身体诞生之前，就和释迦牟尼一样，都是宇宙的智慧先知。他们在地球诞生之前就出现一些光球、白象入胎、鸽子之类的，就是伟大的先知的意识从外星直接传过来的。大家可以看一下我以前讲过的“师尊诞生的秘密”，他曾经就是天界最高的大师叫做净光菩萨，就是从天界一束光直接传到了地球，然后摩耶夫人白象入胎。耶稣也一样，除了天使报喜之外，在1970年荷兰画家阿特德戈尔德画了名画《耶稣受洗》，他就完美记录了一束光给耶稣带来了新的智慧，天空中出现了一个巨大的UFO向地面的耶稣垂直发射出激光之后，耶稣一下就醒了。这个非常有意思，大家要相信光啊！

五角大楼这个档案里还记载，在1963年和1964年，Lin又联系了Bob，还把他带到了宇宙飞船。第一次是进入了一个侦察飞船，第二次侦察飞船就把他带到了距离地球5000英里的太空，进入了一个巨大的母舰。在里面见到了克雷多星球的长老级大师Aklim Vesta Antidi。这个大师就给Bob介绍了一些更高级的概念：第一个就是“宇宙星际联盟”，说在宇宙有一个联盟，包含了银河系内外一共有5000多个文明的星球。他们一直想把地球纳入进去，但是发现地球的科学技术是接近达标了，但是社会与道德水平严重不合格。说地球是一个充满种族歧视、民族国家仇恨、自私自利并且时刻毁灭自我的一个文明，所以绝对不允许这种文明进入一个没有战争的宇宙联盟。如果他们进入的话，肯定会发动战争。所以宇宙人必须等地球人在精神上和宇宙人达到万物一体、全种族皆为兄弟的觉悟才能加入。

之后，大师还给Bob介绍了他们宇宙人的能力，这解释了我之前一个非常大的疑惑：为什么说这个佛啊、观音菩萨会不断的变换自己的身体大小，还能变化出无数的形象在地球上救苦救难？这个大师就说，不管人类还是宇宙人都有心智，也就是意识。而这个意识分为五层：无意识、潜意识、意识、超意识，一直到最高的意识叫做All Mind（全意识）。宇宙里所有的物质、所有的能量、所有生物的心智，在最高层All Mind全连接在一起，成为一个巨大的无形网络。克雷多大师就是说这个宇宙里的巨大意识网就是造物主，就是上帝，万事万物都在这个All Mind里面。他说这个概念就和释迦牟尼讲的“阿赖耶识”，又叫做藏识，都是一样的。他认为是宇宙万法的总根源，一切的现象都从这里面产生。印度教也讲这个事情叫做Brahman（梵），说这个Brahman就是遍布一切、超越一切、万物归一的终极存在。

大师还讲，如果你的意识可以达到All Mind这一层，那就可以将你的物理身体和宇宙最高的能量网络连接在一起，获得无限的知识。并且强大的精神力可以直接下达命令，这个All Mind就会在原地将你的肉体直接转化为纯能量，或者是把身体分解成最基础的原子，然后在宇宙的另外一端，用那里的游离原子瞬间把你重新拼装起来。这个非常玄幻，但这一下子就解释了为什么释迦牟尼说他自己成佛之后就获得了六神通，一下子就连接到了All Mind了。其中一个就是“神足通”，在《摩诃摩耶经》里面记载，释迦牟尼经常这么一个意念，刷了一下就去了忉利天给天人说法，还给自己母亲摩耶夫人讲法。释迦牟尼是怎么去的？估计就是这么一个意念过去。

这个宇宙大师说，要呼唤All Mind达到最高的超意识，必须达到一个特别的宇宙频率。这个频率就像宇宙共享的公共广播一样，宇宙人的大脑非常发达，他们都可以接入这个频率，随意的开启和关闭。当你成功激活这个超意识之后连接到宇宙频率，你的心智和最高层的All Mind直接就产生了共振，这个时候你的网络就已经接通了，就可以像释迦牟尼一样看到过去一切的因缘，这就叫“宿命通”；还有什么看到一切“天眼通”、听到一切“天耳通”，还有瞬间移动“神足通”等等。

大师就说，地球人如果掌握这些随意调频、呼唤全新的这种智能，必须从出生起就开始培养，如果地球人到了五岁以上再去尝试要激发这种能力，就变得非常难了。按照他的理论，说这个世界上也不存在一个像基督教里面那种白胡子老头一样的神。这个神其实就是这个宇宙中的All Mind，宇宙中所有的物质、能量结合在一起的巨大统一体。它记录了所有的过去，但是他们有一个问题，就是他不知道未来，这个未来是动态的。所有的未来都掌握在拥有自由意志的人类和宇宙人的手里。这个All Mind只能向智慧生物提出建议、进行引导和指引，但是它无法强行控制人类或者宇宙人的自由意志。

而且它有一个特点，就是所有拥有自由意志的人类和宇宙人，它都可以反向控制这个All Mind。所以这又解释一个问题：为什么地球中古老的经文都要写做人要向善、要诚实、要和平？因为如果全人类、所有宇宙人都陷入了对战争、恐惧、仇恨这种庞大的负面集体意识，就会驱使这个All Mind将地球甚至宇宙导向毁灭。相反的，如果所有的有意识的生物都渴望和平、渴望博爱，这种集体的念力就能直接改写宇宙的最底层代码，强行把这个太阳系、银河系和整个宇宙的未来通向爱与和平。这个就非常有意思了，可能这就是为什么耶稣、释迦牟尼都说要爱、要和平、不要战争，“God is love, love is God”，神就是爱这种话。

最后这个宇宙大师还给Bob讲了一个非常厉害的月球的历史：月球曾经被毁灭过。他说在几百万年前，月球上面是有完整的大气层、有适宜的温度、有植物也有生物，也有一个非常高等的文明叫做“Lunares”（月球文明）。他们都城就在哥白尼环形山里面。但是几百万年前的一天，有一群未知来源的更高级的远古外星种族降临了（是不是阿努纳奇异族不知道）。一开始这个月球人还把他们当作神，修建了巨大神庙给这些外星人做基地，但后来月球人的科技不断发展，慢慢发现这种远古的外星种族也不过如此，也不是什么神族。于是就想让他们赶紧离开，“滚蛋吧”。但是这个远古外星人就不离去，双方就产生了分歧，结果就发生了大战，而且是核战。不知道哪一方用了某种粒子核弹头，瞬间就把这个首都炸到气化。这种能量强大到它把大气中的所有原子直接转化为纯能量，于是整个月球就从他们首都哥白尼环形山这个地方，逐点形成一组巨大能量的狂怒火墙。它像巨大的涟漪一样从爆炸中心慢慢扩散，不到一分钟的时间，整个月球就从绿色的、有生机的星球一下子就变成了死寂的灰色废墟。按照他的说法，这个月球和地球当年都是绿色的，都有生物存在。

这非常有意思，因为按照日本最早的物语文学《竹取物语》来说，他们就记载了一个从月球下来的人叫做“辉夜姬”。而这个辉夜姬，她就来自“月之都”。当时日本天皇还拜访辉夜姬，他们之间相爱，天皇向她求婚，但是辉夜姬拒绝了，说“我不是你们国家的人，我来自月之都，必须返回月球返回我的族人身边”。之后她就准备和人类分别，于是大家就看到了天上有飞行物下来，然后带着辉夜姬刷一下升到了天空中。他们升空的地点就是现在的富士山，富士山在日语中念作Fujisan，它来自另外一个发音叫做Fuji，在日语里这个是不死的意思。辉夜姬离开的时候就留下了一种药，叫做长生不老药、不死药。这个故事也非常像嫦娥的故事，嫦娥好像也是跟不死药有关。Fuji这个就非常有意思了。

最后五角大楼大案里面就记载，洛杉矶分局探员听完之后就面面相觑，不知道发生了什么，怎么这么玄幻。所以在报告中他们就总结：Bob Reynolds在之前从未参与过任何被官方认为激进、颠覆、间谍或犯罪的活动，他是一个普通的美国公民。FBI洛杉矶办事处在1966年10月3日呈报给局长约翰·埃德加·胡佛，在备忘录中直接下达了结论：不对此事采取进一步行动。

这个就是今天的故事，非常玄幻，竟然可以在五角大楼档案里面出现，真的是非常不敢想象。也许我们的宇宙就是这样，我们真的是要好好修炼提高自己的意识啊，打开阿赖耶识，最终和宇宙的All Mind连接起来。说不定现在的释迦牟尼就在这个All Mind的里面。我现在真的是觉得这个佛经里面讲的非常非常有道理，大家都好好研究一下。

If I were to right now and graduating college, I would feel like the luckiest kid in all of history. Why?

Wake up. AI is here. And if you assume any rate of improvement over any reasonable time period, learning how to use AI should become your number one priority. I think it is possible now to start a company that is a oneperson company that will go on to be worth more than a billion dollars. The New York Times just ran a story about a guy who built a billion dollar company with AI. We all knew it would happen. One person creating a $1 billion company. You have access to tools that can let you do what used to take teams of hundreds. And you just have to learn how to use these tools at home with a great idea. You probably heard it about a thousand times. Oh, AI is going to make you rich. Oh, I just made Claude code wipe my or I just replaced my entire family with Claw Code. There’s a bunch of those videos out there and they have taken the internet by storm. And I know what you’re thinking. No, another one of these drop shipping NFT get-richqu schemes. But that is not what this course is about. And if you’re looking for the next get rich quick thing, then you might as well just click off this video cuz this video is not for you. And I know that because I actually thought the exact same when I got into this space a little over 3 years ago. If you don’t know me, my name is Elbert and I am by no means the smartest, but I still somehow managed to build two AI companies that combined have done over a million dollars, which is weird cuz I don’t have a university degree in computer science. I actually never went to university. So, I’m kind of on my what, like seventh gap year or something. The only reason that this was possible was because of AI. And I give full credit to Chad GBT and Claude like 100%. I couldn’t have done that without these tools. But let me show you how this was even possible. You might have seen this graph going around. Each dot on this graph represents 3.2 million people. So in total, we have 2,500 dots, which amounts to 8.1 billion humans. And what this shows is how many people in 2026 that have even used AI. And what you will find is that 84% of people, 84% which is 6.8 8 billion people have never used AI in their life. They have never even given chativity a regular prompt, which is hard to believe because if you’re like me, you see AI everywhere. Every time you open your phone, you hear AI news, but that is because your algorithm knows that you’re interested in that stuff. 16% of people have used free chatbots. So that is free chat GBT or free claude. That is 1.3 billion people. But if you are just a little into AI, you know that the free models, you almost can’t do anything with those. And then if we zoom in, we have the people that pays $20 a month for AI. That is 25 million people represented by these small yellow squares right here. 0.3% of the population. And my guess is that’s probably where you are right now. Or maybe you are part of the 0.04% which is the max users. So the people using the most powerful models like claw code and codeex that is only 3.6 million people or 0.04%. The reason that I’m showing you this is that I want to show you how far behind the world still is when it comes to AI. Many say that AI is going to be the greatest productivity boost that the world has ever seen. And even then after AI has existed for a couple of years now, 84% of people has still never even used it. I hope you see how this creates a massive opportunity because this allows us, the 0.04%, to provide services to the rest of the 8 billion people that don’t know how to use AI effectively yet. To these people, your AI services is going to seem like magic. And the reason that I’m dropping this full course right now is that the world is finally starting to understand AI. When I started providing AI services 3 years ago, the only real markets that we could go after was the US and Canada, the UK and Australia, every other market simply wasn’t ready yet. But because of the mass adoption that’s happening right now in AI, all of a sudden most other countries are starting to pick up as well. And this opens up markets like South America. Europe is one of the biggest upand cominging markets in AI right now. Countries in Asia like India is also starting to pick up on AI which means that companies are ready and they know that they need AI implemented into their business and it’s not going to take long before the entire world is at the same wavelength. You might be thinking well AI companies already implementing AI I might be too late but I showed this to you to make you understand that it’s not too late yet. And by the end of this course when you stick around you will 100% be in the 0.04% 04% of AI users that will actually be able to capitalize on AI. But don’t get too excited yet because if you’ve ever tried to start a business before, you might know or definitely have felt the excitement curve before. When you’ve just heard about a new business model or you want to start something or a new project or a new business, you’re going to be extremely excited. Your excitement is going to be through the roof. You’re going to be thinking about all of the good things. So, for example, with building an AI business, this would be, “Oh, I can automate my entire product delivery. Oh, I can get unlimited clients. Everyone’s going to be interested in my services. I’m going to be rich.” That’s the thoughts that are going through your head when you’re just starting out. And the reason that I told you right at the start that you shouldn’t expect this to be a get-rich thing is because your excitement will then drop. This is 100%. All of a sudden, you realize all of the problems that are in the business, that it’s actually not easy to get your first client, that you have to get really, really good at what you do. You need to watch videos like this. You need to learn before you can actually start selling these services, and that you probably won’t land your first client the couple of weeks after starting. That is the reality of every single business model. There are always issues. There are always things and things are never as easy as they seem. So you reach a low of excitement. And this right here is where 95% of people give up. They don’t want to go through the pain of figuring out, okay, how do I actually land clients? They don’t want to spend all of the time learning the skills that it actually takes in order to provide good AI services. So they do one of two things. Either they say, okay, this AI thing, scam, onto the next. They try dropshipping or NFTts or crypto or some other thing. or they simply just stop business altogether and they go back and just focus on the regular thing that they’re doing the 9 to5 or whatever but a small percent of people 5% or so are going to work through the excitement drop and it’s not a steep curve it takes time and all of a sudden they fix a couple of issues and then they land their first client so it might go up like this but then they realize okay it’s actually not as easy as I thought providing the service so it stalls again and then they maybe figure out okay how do I actually provide at the service. So, it goes up a bit again and then their first client drops and all of a sudden they’re back with no clients. But as they keep working, this curve starts going up and all of a sudden they crack how it works. And after long enough time, they’re going to be very excited. They’re going to know, okay, this is how I actually scale this business. This is how I do it. But getting to here takes an immense amount of time and work. I’m setting these expectations right now because if you’re not ready to do this, then it doesn’t make sense for you to spend hours watching this. you can just click off the video. But if you’re actually committed to do this, if you are the top 5% that does to give up, then you are the person that I made this video for. The worst thing you can do is to get stuck in this loop of trying a new business model, getting very excited, getting the drop in excitement when you actually try it and say, “Okay, this doesn’t work.” And go on to the next one. because all of a sudden you get stuck in a loop of just trying stuff but never actually committing and going all in and actually staying with it, which is what it takes to get success in anything in life. I hope you’re ready to put in the work. And if you are, then good. Listen up. These are the things that we’re going to go through in this full course. Don’t worry if you are a beginner in all of this claw stuff. We’re going to start all the way from scratch designing and building our AI operating system from nothing. So, if you’re just starting out, this is the perfect video for you. Most other guides on YouTube show you how to build the tech, but they actually never show you how to land your first client. So, that’s actually the first thing that we’re going to do in this course. We’re going to focus on getting clients, getting clients, getting clients, because that is really what matters and really what moves the needle when you’re just starting your AI business. The best way to land clients when you’re just a beginner is to do something where you can reach out to a lot of people, but do it very cheaply. So, I’m going to show you how you can set up cold email campaigns, and I’m going to show you how you sign up to Upwork to get your first couple of freelance clients. Then, I’m going to show you how you take those interested leads that we got from cold email and Upwork, and how you book them in on meetings and how you close them. And I’m even going to show you a secret trick for how we can get Claw to listen in on our meetings so it gets the context for exactly what we need to build and can even start the building process before we even end the meeting. So, stay around for that cuz that’s absolutely going to blow your mind. Then I’m going to show you how you actually do the service delivery. I’m going to show you how to build websites, automations, mini apps, and the process that I use for basically being able to build anything with no university degree where Claude Code does the heavy lifting for us. And then we’re going to go over a very important step which a lot of AI business owners, they miss. We’re going to take the cool things that we have built and we’re going to show them to the world. We’re going to post this on LinkedIn. We’re going to post it to YouTube. And this will have a compounding effect where people all of a sudden see the proof of the stuff that you have built which is going to get a ball rolling to get you even more clients. And at the end I’m going to show you how not only to automate your clients businesses but also how you automate your own business so you truly achieve the one person AI business using Claw Code. This is probably going to be the most valuable thing that I’ve ever released and I’m releasing it for free. So I’m very excited. Let’s get into it. The beauty of Claude Code is that we can manage our entire business with it. So, let’s set that up, shall we? If you have already set up plot code inside of Visual Studio Code, then you can just skip to the next chapter of the video. But let’s set up Plot Code the right way. The first thing you want to do is to open a browser. Then you want to search for Visual Studio Code, click the top link, and then download it for your computer, either Mac or Windows. Visual Studio Code is what’s called a code editor. But don’t worry, we basically not going to create any code ourself. CL code is going to do absolutely everything for us. We then click on the installer. Then we drag it over to the applications. We let it install. And now we can open up Visual Studio Code. You’re going to land on this page right here. Before we do anything inside of here, let’s create a folder that we can work within. Damn. Excuse the mess on my desktop. I promise you we will get that cleaned up. I’m going to create a new folder and I’m going to call it shiny which is the name of our agency. This is going to be where our AI operating system lives inside of this folder. So call it something that you can remember. Then head back under Visual Studio Code and click this open button right here. Then go to wherever you saved the folder. I saved it on my desktop. So I’m going to open this shiny folder. Click open. And there we go. We’ll then click yes. I trust the authors. and we can close down this welcome message. Now we are inside of our code editor and we are ready to start setting up clawed code. What you want to do is that you want to open the terminal. So I’m going to write click inside of this space and click new terminal. And this is going to open up this terminal view right here. Don’t worry, this is going to look a bit technical, but you can just watch exactly what I do onetoone and it’s really not going to be that hard. You don’t need to be that technical for this. Then I’m going to open a new tab. I’m going to search for claude code install. Click the first one that’s called quick start cla code docs. And then we have three commands right here for installing it on Mac or on Windows. So I’m going to choose the Mac install that fits with my computer. Then I’m going to paste in this command and that’s going to install cloud code on our computer. And there we go. It now says setting up Claude Code. Installing Claude Code native build latest. It then says Claude Code successfully installed. It shows the version. It shows where it was installed. So now we can write clear. That’s going to clear our terminal. When we now write cla, you can see it’s going to say welcome to clawed code. And we can choose what style we want. The way that we navigate is with the arrow keys up and down. And I’m definitely going to want dark mode. So I’m just going to hit enter to dark mode. And now it gives us three options. Either a cloud account with a subscription or using an API key through the console. I’ll definitely recommend you use claw with a subscription. either pro or max. I’m personally on the max plan right now because I need that bit more usage. But if you’re just starting out, then I recommend that you start on the pro plan. So just go to claude.ai, sign up to the pro plan, and you can see that gives you access to Claude code directly in your codebase. When you’ve signed up with an account and you have the pro plan, you now go to Claude account with subscription and you hit enter. And this is going to ask you to authorize with your Claude account. You can see it says Claude code would like to connect to your Claude chat account. So, we just click authorize. Now, it says build something great. How fitting. And you can see it says login successful. Press enter to continue. So, that’s what we’re going to do. We’re going to click enter. And then we’re going to hit enter again. And then claude is asking us if we want to use recommended settings. I’m going to hit enter. And there we go. Now, we’re inside of cloud code. And then we can write just claude like this in our terminal after we’ve installed it. That’s going to open up Claude code on our computer. It says quick safety check. Is this a project you created or one you trust? So, I’m going to just hit enter. The way that you navigate inside of these options is that you can use your arrow keys up and down to go up and down. I’m going to click yes, I trust the author, and then hit enter. And there we go. Now, we are inside our claw code view. And we have successfully installed clawed code. The next thing we’re going to need is some clawed skills. Inside of our completely free group, you can grab my clawed skills. I’ll leave a link right below this video. And then you can find those under classroom and then go into AI learning hub. Scroll down until you find claw skills right here. And this gives us this drive folder with all of our claw skills. So I’m going to click claw skills right here. I’m going to click download. And now you can see it’s downloading all of these skills in a folder. Claude skills are basically portable skills that you can give to claude code. So inside of these skills, we have some documents that explains how to use something. It’s basically just a list of instructions. But by having the right skills and having the right instructions, you can make claude code even smarter than it already is. So if we go back into cloud code and we then doubleclick on the zip file that we just created, this gives us our claude skills. I’m now going to write exactly this inside of my folder Claude Skills 2. This was the folder name, right? The folder name is Claude Skills 2 in my downloads because it’s inside of our downloads folder on our computer. I want you to install these skills in this project folder. And then I’m going to call the folder shiny where we are right now. Then we’re going to write exactly this. I want you to research the documentation for installing claude skills in a project. Please research that and install these skills in this current folder. Hit enter. And now what we’re basically forcing Claude to do is that we’re forcing it to go out and research first before we have it do anything. And that is also how you usually work with Claude. Sometimes it doesn’t have all of the context that it needs. So by telling it to go out and find that context, we’re going to get much better results. That took about well exactly 43 seconds. And if we go inside of thisclaw folder that was now created on the left hand side and inside of the skills folder, we can now see all of the skills that we have inside of this folder. What you want to do now is that you want to write clear inside of cloud code. Then we can hit Ctrl C, which is basically just going to stop cloud code. And then we’re going to force it to restart. After installing skills, you need to restart the session for the skills to actually be enabled. If I now write slash and then front end design, you can now see we have a front-end design skill right here that creates distinctive production grade front-end interfaces with high design quality. Use this skill when the user were asked to build web components, pages or applications blah blah blah. So now we have a front-end design skill and we have all of these other skills as well like composio or customer support or research or scalability or security etc. All of these skills are now inside of this. Claude folder inside of our workspace. The next thing you want to do is that you want to be able to save this to the cloud. Right now, all of this information, all of these skills are living on our computer, but we want to have a copy of this in the cloud as well. That way, if I smashed my computer and for some reason it was destroyed, we would still be able to recover our workspace, which if you have been working in a workspace for, let’s say, a couple of weeks, you definitely don’t want to lose that workspace and all of those skills. Let me show you how to do that. Go back into your browser, then search for github.com, and create a user if you don’t have one. After you’ve created the user, you’re going to land on this page right here. Then you want to click new. And I’m just going to call this shiny workspace. I’m going to choose myself as the owner. And then I’m going to make it not public because we don’t want everyone to be able to see our workspace. That is a high value asset. So I’m going to choose private. And then click create repository. This will basically create a folder inside of GitHub. And with GitHub, we can basically publish our code that allows us to save a version of it in the cloud. The next thing you want to do is that you want to copy this URL. Paste this in and say, “Please push this current folder we are in to this repo.” And now what Claude is going to do is that it’s going to take these claw skills, which is everything we have inside of the repo yet. And it’s going to take all of that and push it into the cloud. So again, if something happened to our computer, if we lost access to it, we don’t lose our workspace. Very important. My GitHub account is already connected to my computer. When you’re doing this for the first time, it’s going to pop up and ask you to authorize with your GitHub account. So, just make sure to do that when you’re doing this for yourself. You can now see it says the push was blocked by an auto mode classifier. That’s because we have clawed in auto mode right now. What I’ll do is that I’m going to set it to just like default mode, right? Try again. And then we just have to accept that it actually takes this workspace and pushes it into the cloud. Do we want to proceed? I’m going to go down and say yes and don’t ask again. Hit enter. And this will now be pushed into the cloud. pushed successfully to Alberta shiny shiny workspace main branch 52 files. If we go back into GitHub, you can see this is what a blank workspace looks like. If you now hit a refresh, we’ll be able to see our claude skills right here. This is the exact same files that are inside of this folder right here. Now, they’re just also in the cloud. Great. Congratulations on having all of this set up. Now comes probably the most valuable part of this entire course, and it’s really what no one else is showing you on YouTube. Everyone is talking about the tech, but the hard part for almost everyone is, okay, now I know how to build stuff. How do I actually sell it? How do I go out and actually land clients? A wise man once said that you have to get good before you can get rich. What do I mean with that? What you’ve probably seen on Instagram and all the other socials as well is people saying, “Go out, learn how to build websites, and sell your first one for $10,000.” And while it is true that you can sell websites for this much, is it realistic that you’re going to go out and close a 10K deal as your first one? Well, probably not. And that is also what I mentioned earlier when we talked about expectations. It’s probably not going to go that way. Why is that? Well, there’s a couple of reasons. The first one is that you have absolutely no trust. And how do you build trust? Well, you need testimonials and you need previous experience. So when you hop on a call with someone that you actually know what you’re talking about that you’re actually certain in yourself that okay you can help this person get better results. In this course I’ll show you how to get both of those things. And what I’ll also show you is the next part that you’re lacking which is skills. For some reason these days people think that you can just go out and learn something in 5 minutes and all of a sudden charge thousands of dollars for it. I’m sorry but that is not reality. You’ve got to understand that what people are paying you for is the time and effort you’ve put in to gather skills. Skills that this person that are paying you doesn’t have themselves. If it was easy, the thing that you’re doing, the thing that you had learned, then no one would pay you for it. So, you have to do exactly what it is that you’re doing right now when watching this video, you have to go out and you have to learn, you have to build these skills in order to get paid. And that’s exactly what I mean when I say you have to get good before you get rich. And of course, that’s also the entire point of this video is to get you to build those skills. When I started my AI business, it didn’t take a week. It didn’t take a month. It didn’t take two months. It took four months before I closed my first paying client for $400. And guess what I did before that? I did not charge $10,000 for a website. I actually did free work in order to actually build trust, to gain experience, and to gain testimonials. And by doing the free work, I also gained the skills that I could later monetize and actually charge for. So if you’re thinking right now, okay, I’m going to go out, I’m going to learn this, and then the next week I’m going to sell a 10K website, leave this video. This video is not for you. Understand that you have to build the skills before you can charge anything. And understand that this takes months. The way that I think about it is that there’s kind of like three different levels to what you can be building when you’re just starting out. The first level that I would recommend that you start with is AI generated websites, specifically with cloud code. It’s extremely good at building websites. And let me show you something very, very interesting. If I head over to Claude and I ask how many small businesses still don’t have a website, while I then drink my coffee, Claude is now going to research the web and find out the number. And now we can see that roughly 27 to 30% of businesses in 2026 have no website. So roughly one in three small businesses still don’t have a website. So if you learn that skill of building websites, all of a sudden you can provide services to that 30%. And that 30% does not have a website. You can also build better websites and sell it to the rest of the 70% if you provide something that’s better than what they currently have. So I can write, okay, so if it’s 30%, how many businesses in that? in let’s say the US just to put a number for how many potential businesses that you could actually be working with. I’ll take another sip of the coffee while Claude is researching. So in the US there are 36 million small businesses and 30% of that is then 10 million businesses without a website and that is in my opinion the best place to start. Now you have millions of businesses that you could service with your skills. They don’t know how to build websites with something like cloud code. They are not going to learn it. They are focused on their craft. So when you in this course is going to learn how to build websites, all of a sudden you can service all of those businesses and provide a valuable service. That is level one out of free. The next thing we’re going to take a look at is the step up. Now you’re not only building websites, you’re also automating their business with automations and agents. This requires some more knowledge into integrations and APIs, but don’t worry, you’re going to be learning how to do that as well. And I would say the final level is building out full AI systems for businesses that not only integrate into what they already have, but also where you are collecting all of the automations that you have built from level two and then maybe creating a dashboard or somehow building it into a full AI system. But of course, we don’t start at level three. We start at level one, websites. And stay around because later in the video, I’m going to show you exactly how you build those out in just a couple of minutes. I wanted to go over these three levels so you know what to actually sell because in the next part of the course, we’re going to go into, okay, how do I actually land clients? One of the biggest mistakes I see beginners make is that they focus all of their time on building and not enough time actually outreaching and talking to clients. We’re going to flip that on its head. So, before you even know how to build anything, we’re going to start with the outreach because that means we can have the outreach running while we are then building stuff. So, let’s get into probably the most valuable part of the course, how to actually get clients. Let’s get into it. to do our outreach and actually build up the marketing system. We will of course also be using Claude Code. We’re basically going to be using Cloud Code throughout the entire video and Claude Code will probably do 90 to 95% of the work in this business model that we are building up right now. I’ll be showing a bunch of cool stuff that you can do inside of Claude Code. If you’re looking for a full guide to Claude Code, then I’ll definitely recommend that you check out the Claude Code masterass inside of our close community, the 1% in AI. This is also where you can get help directly from me. You have a full master class in absolutely everything inside of Claude Code that takes you from a complete beginner to a master in AI coding models. This community is also where you can get help from me. I answer every single post inside of this community. If I click on these icons right here and then go to new, then you can see that I answer absolutely every single post in here. This right here is my face and you can see I have commented and answered every single post inside of this community. I can actually see a post was just made 1 minute ago and just now. So, let me answer those right quick. That was one. Let me answer this as well. There we go. And the best thing about this community is that right now we are running the 30-day AI challenge. What this is is basically 30 days of videos. And when you finish those videos and stay consistent for 30 days, then you actually get your first month completely refunded. So, if you’re consistent, you can start the 30-day AI challenge that takes you from a complete beginner to setting up an agency in 30 days. And then you can get to try all of this completely for free because when you finish, then you get all of your money back. If that sounds interesting, then watch this video right here. This explains everything. I’ll leave a link right below this video. All right, let’s continue with the course. Let me show you how you can actually get clients. All right, let’s get into it. This is how I’d recommend that you get your first couple of clients and how you build trust and how you build testimonials. The first ones are always the hardest. It’s like rolling a big stone. It’s going to require a lot of force when you’re just starting out. But as soon as you get the ball rolling, then you can use the momentum and it will start rolling faster. It’s the exact same when it comes to landing clients and building the business. The first couple of months, the first couple of clients are an absolute pain. But let me show you how I recommend that you get your first couple of clients. To do this, we are going to use Upwork. If you don’t know what Upwork is, Upwork is basically a freelancing platform that allows you to post yourself as a freelancer to say, “Okay, I can help with these things.” And then businesses will also go to the platform and say I need help with these things. So this platform basically connects freelancers with actual businesses. The good thing about Upwork is that you can find leads directly that need your exact services. So if you’re building websites then you can find people that need something improved on their website. If you’re doing AI automation system or go high level or inn you can find people that are specifically looking for that service. This doesn’t mean that Upwork is just like free clients and that you’re going to get a client as soon as you sign up to Upwork. There are really levels in the game when it comes to Upwork. How you build your profile, how you send your applications to post, but I’ll cover all of that in this module. So stay around because there’s a high chance that this I’m about to tell you right now will get you your first couple of clients. First thing you want to do, go to upwork.com, click the top right corner, and click sign up and then sign up as a freelancer. I’m then going to use my Google account to sign up. There we go. I write in my first name. I write in my last name. And then I choose the country, which is going to be Denmark. Then I’m going to check this box right here and click create my account. And there we go. Congratulations. You have been signed up. Now we need to create our profile. So I’m going to click get started. I recommend that you write I have some experience right here. Click next. What’s your biggest goal for freelancing to earn my main income? Click next. Then check both these on and check this right here and click next. Create a profile. Here you can do a couple of different things. If you have a LinkedIn that already now shows your experience, then you can connect your LinkedIn already. So that’s what I’m going to do. I’m going to click import from LinkedIn. Then I will go to LinkedIn. Go to my profile. Click these three dots and click save to PDF. That basically creates a PDF of our LinkedIn profile. Then I can click upload. You save LinkedIn PDF. Choose this. And this will now upload our PDF. So Upwork gets all of our information from our LinkedIn. Don’t worry if you are not set up on LinkedIn yet. If not, just click fill out manually right here and go through the process. But for now, I’m going to click continue editing your profile. Now you’re choosing what kind of work we’re in. I’m going to go inside of IT and network. and then choose database management and administration and CRM software right here. This is what best fits like a classic automation agency. Now we click next. Add your skills. Now we need to add a couple of skills. We could add CRM software already here. We could add our platform that we’re working like in for example. We could add cloud. We could also add things like marketing. If you’re doing Facebook ads for example, you could choose Facebook advertising. You could choose things like Google ads. You can choose things like building websites. You can always change this later, but for now it’s good to show, okay, these are the things that you are usually working with. You can also choose AI agent development and sometimes it’s also good to choose some coding languages like TypeScript, Python, etc. That can sometimes help you get some jobs. You could also choose things like Savia for example. This can help us as well. So basically like the automation platforms also choose make.com. And there we go. Now we have added a bunch of skills that shows what we can offer. We can then click next and add a profile title in our professional role. We can write something like this. AI agents and savior go high level make in it in AI consulting and implementation. Then click next. Add your experience. You can see it has found that I’m co-founder of shiny AI solutions. You can click edit right here and then we can add more information like the location for example is in Denmark. Click I’m currently working in this role. Click save right here. Now we can add our education. Then you can add it right here. For now, we’re not going to do that. Now we’re going to click skip for now and then add languages. I’ll definitely recommend that for languages that you choose native or bilingual right here. You then choose all of the languages that you know. I’m load Danish for example. I know a bit of German and I know a bit of Swedish as well. I can then write next. Write an overview. We’re going to come back to this later. I actually already like the this AI version right here. As a co-founder with four years of experience in tech landscape. We can change this in AI automation and AI agents. I specialize in developing solutions that drive efficiency and productivity and solve real world problems. And then we end it off with if you need someone who is an expert in automation. I am your guy. This one I probably want to leave. You don’t want too big of a description. I like this one right here. It’s also a good idea to write in the tools that you use. So I specialize in developing. I’ll change that to building. I specialize in building solutions. specialized in building helpful solutions that also made it improve real world problems. I use tools such as go high level savior make inn like this if need someone who’s an expert in automation I’m your guy we can even also write claw code I use tools such as go high level savior make inn and plot code /c if you need someone who’s an expert in automation I am your guy there we go now we go next set your rate I’d recommend not charging too much in the start what you’ll see the expert on Upwork charge is something around like $50 an hour. For now, you just want to go out and land your first couple of jobs because those are going to show the success rate is going to show on your profile. So, I recommend that you set this to something like $15 an hour. The service fee, you cannot change that, which means you will get $135 more. Now, we click next. Add your photo and location. For the profile photo, you want to add a professional photo of yourself like this one right here. So, I’m going to click upload photo and then upload this photo of ourselves right here. Click attach photo. Then you want to write in your birthday, your country. Fill out all of this information and then review our profile. There we go. Agents and Savior go high level make and then consulting and implementation. Set our hourly rate, set our skills, set our work history, and then submit the profile. Now, Upwork is asking us to buy some connects. I’ll definitely recommend that we do do that on one point. For now, I’m going to click browse without bidding. And we’re going to go to this page right here. And you can see we are 60% complete. Now, we still want to add a bunch of stuff to our Upwork profile. It’s important that you fill all of this information out as much as possible because this helps build trust with Upwork, which will help you land more jobs. So, inside of profile settings, I’m going to set myself as an expert in the space. We can link our accounts. Like, if you want to link our GitHub, we could do that. Going to click authorize. There we go. Then, I recommend that you complete this working style right here, which is basically just a small quiz. I click start assessment. Just fill out these questions with what you believe. stay organized. I like to communicate my schedule. When scheduling my week, I usually plan detailed steps, block time. When I deliver work to a new client for the first time, I usually double check every part before sending it out. When timelines are tight, I prefer to focus on the key task that ensures quality. Before submitting work, I usually review the brief. I confirm it meets the client’s intent. When collaborating with other freelancers, I usually cross check shared work for consistency. When project deadlines change suddenly, I talk with clients to confirm what’s most important to finish first. I feel most productive. When I’m creating solutions, when I’m faced with something new, I usually start by brainstorming. When I need to make a decision, I typically make the call and take responsibility. Ling the golden key so I can make like this. Before I start to work with a new client, I prefer to identify potential challenges early and plan the way to handle. There we go. Clear communicator is now visible on our profile. All of these things basically just helps you build trust with both Upwork and people looking at your profile and all of these are really important. The next thing you want to do is verify your identity. You do that by clicking this button right next to your name. Click on portfolio and then you can add a project title like a solar sales AI agent that I’ve done for example reactivation of old solar leads. This can literally be anything. Also, if you had a previous job, write in what you did and what you accomplished. Then I have this picture right here from our website. Let me take a screenshot of that. Now we can add images. I also have a picture right here of a completed booked calendar that I write done. I can write book calendar book by an AI agent. I can add more images like this one booked in a month. Next preview. And there we have it. So my sales agent developeration leads the next thumbnail. Let’s choose this as the thumbnail cuz that’s probably the most catching picture. And there we go. Our portfolio is published. Now you can see we are 75% done. We could also create a video introduction if you wanted to. We could add certifications. If you have any certifications that you’ve done that are slightly similar to what it is that you’re offering now, make sure to show that on your Upwork profile as well. Including an employment history is also smart. So, if you have a previous job, include that, too. Basically, fill out as much information on this Upwork profile that you can. One of the things that can make or break your Upwork profile as well is the location right here. So, you want to make sure to go inside of profile settings, go inside of contact info, and then change this to a location that people know. Right now, I’ve set it to a city called for expand Denmark. I’m going to get better results if I set it to something that people know like Copenhagen. Then, we can click update. And now, you can see it says Copenhagen, Denmark, and people know what Copenhagen is, which one sparks new conversations, and two, builds credibility, which is what all of this is about. But we can improve this up profile a whole lot more. And of course, we need to use cloud for that. Let’s get into it. Before we continue, let’s take a look at the people that are absolutely crushing it on Upwork. Like this guy, for example, Vipa D. Total earnings a million on Upwork, 2,000 hours worked. And if we take a look at his profile, we can see it says, Turkey, you know, a capital big city that people know. 98% job success and top rated. And the great thing about these Upwork profiles is that they are 100% public. So, we can just go in and steal what we want. You can see he’s offering a consultation first and he’s showing some work history with his reviews. You can see that he has a bunch of languages right here that he’s showing. He’s ID verified. I’ll show you how to do that later as well. It’s showing education and then showing a bunch of reviews. It’s showing his portfolio of things that he’s already worked on. It’s showing certifications, employment history. You can see he has a very stacked Upwork profile. The most important thing for an Upwork profile is this job success right here. And also this text right here that says available now for the job success. We literally just need to get some jobs done. And that also means that we don’t go after the $75 an hour jobs just yet. If we can get a job for $15 an hour, even if it doesn’t make us much money, it’s still a massive win because it’s going to show on our Upwork profile. Let’s move on to this guy. You can see also available now 100% job success. He’s charging $62 an hour. He also has a booger consultation. So there’s definitely a pattern here. He has worked 4,000 hours in total. He has perfect reviews, a portfolio, a bunch of skills, and a project catalog. Same with this guy. Also available now. 100% job success rate, $57.63 an hour, 200k earned, also book consultation. So, we should probably add this. And then a bunch of work history, and a bunch of things that he has already worked on. It’s always important when you’re starting something new that you look at the people that are actually crushing it to steal all you can. What I’ve done is that I’ve actually taken a bunch of these high performing Upwork profiles. I’ve given them to Claude and created a Claude skill which means that we can literally just pass in our own Upwork profile and Claude can improve it. Let’s do that. So now we can head back into Claude code. The Upwork skill should already be installed for you from all of my skills. Just write slash upwork and then we want to go to our own Upwork profile. Basically just Ctrl+ A to copy everything and say improve my Upwork profile. Paste this in. And because of the Upwork skill that is already trained in the best performing profiles in our space, CL can now take that as a reference and then improve our Upwork profile. Of course, your Upwork profile should be with whatever you want to sell. I’m selling AI lead follow-up systems. So, I’m going to copy this one. Actually, I kind of like this one. AI automation engineer. Let’s change this. Insert one of these as well. AI automation engineer lead follow-up systems and go high level. Instead of only lead follow-up systems, we can write AI agent systems and go high level and make. Click save. Now you can see that claude is auditing our description as well. This is probably pretty good. I like this. So let’s copy that and insert it.

Click save. This is a pretty good call to action. Message me with whatever it’s leaking. So let’s include that at the end. I like this. And I’ll tell you in 24 hours if I can fix it. That’s like kind of a call to action on our Upwork profile. So we can write message me with your project and I’ll tell you in 24 hours if I can help. Great. I think our Upwork profile is looking a lot better. Another tip that you want right now I just set the price to $15 an hour. You want to be more precise. Like let’s set it to $16.73 an hour for example and click save. The reason for that is that if you just set it to let’s say $16 straight, it doesn’t seem like a precise price. It’s just like thrown out there. When you set it to something very precise, it makes the human brain think, “Okay, there’s a reason that it’s this precise, which can build credibility.” Awesome. Our profile is starting to look a lot better with the help from Claude. Now, it’s time to sign up to some jobs. Before we do that, you need to know about the two plans inside of Upwork. Right now, we are on the basic plan, right, which is the free plan. And here, we get 10 connects per month when we finish our Upwork profile, which if you are serious about building out your Upwork profile, probably won’t be enough. They have an offer right here though. You can see the plus plan right now is $20 a month. But if you go to your profile, then you can actually get 50% off right here. Get freelancer plus for 50% off for one month. So that’s $10 for to try it for one month where you get a 100 connects instead, which is very very worth it. So I’m going to go and click upgrade now to the plus plan. Let’s get started with that. And there we go. You can now see we are on the freelancer plus plan. And the first thing you want to do is that you actually want to have back to your profile. And then you want to verify your identity. This was something that we didn’t have before with the standard plan. And this basically just builds trust with both the Upwork and also with our clients. And this also cost 35 connects right here. Right now we have 150. So that’s definitely worth getting this done. So I’m going to click verify identity. And now we scan it with our mobile device. And then we have to verify with some ID. I’ll be back when I’ve done that. And there we go. That took about 5 minutes. So now you can see right next to my name, we now have a verified badge that says this freelancer’s identity has been verified through a government ID check and a visual verification. Nice. On the left hand side of your profile right here, you also have the availability badge. That’s the thing I showed you before where it says available now, which cost 14 connects per week. I would also say that that’s very worth it. So I’m going to click turn on. You’ve added the badge and now we can go to our public profile. We can see how it looks. You can see now it says this available now automation engineer $1673 an hour. There we go. The skills, the employment history, and the only thing that our profile is now missing is some jobs cuz that is really what matters. The thing that people really take a look at is this right here. Top rated plus and 98% job success rate. And then also total earnings, total jobs, total hours. And that’s also why we set our price very low compared to these other people. You can see they’re charging $62 an hour. We are only charging 1/4 of that. It’s because we want to get jobs first so we can have 100% job success rate. Nice. It’s time to go out and actually start bidding on jobs, which is what we’re going to be using these connects for. So, let’s get into that. All right, let’s start actually applying to jobs. If you go to the find work tab right here and click find work, then it’s going to show you the jobs that Upwork thinks that you will be interested in. We can see our connects right here. And the way that connects work is that we are basically paying a little bit every single time that we apply to a job. What does this mean? Well, it means that we need to be cautious with what we apply to. We cannot just apply to every single job. There’s a couple of rules of funds. Like if a post, for example, has 50 plus proposals, it’s probably not the best job for you to reach out to. 20 to 50 is in a much much better range. And it’s really a speed to lead game. When someone makes a post, you want to be instant there. You can see this was posted yesterday. Posted yesterday. You want to be there instantly. This one, for example, right here was posted 3 hours ago and the proposals are 5 to 10. This one would be a great one to reach out to. So, what you want to do is that you want to commandclick on this. So, it opens in a new tab. And you can see this opens up the job right here. We’re looking for a skilled AI automation specialist who has hands-on experience with go high level. That’s us, Cloud AI, and AI agent development to help streamline business operations, improve client communication workflows. There we go. This seems awesome. Remote, freelance, potential, long-term opportunity. There we go. There we go. This one seems like a really, really good job. And you see they are payment verified and phone number verified as well. And they have a 98% higher rate with 910 jobs posted. And again, what we’re trying to do now is not to make the most amount of money. We need to build up our Upwork profile before we can do that. So, if we think we can do this, which we can, we can click apply now. You can see that this proposal requires 11 connects. And when we submit it, we have 103 left. How do we want to be paid? I’m going to choose by milestone. And then we can choose maybe something like 25 25.

And we can do initial audit implementation.

Verify that everything runs smoothly. And you can see when we do this, we are going to be making $67 like this. How long will the project take? Probably less than one month. And then right here, we can add our cover letter. And this is what is important to add. This is basically our application. What I’ve done is that I’ve taken some of the best proposals that I’ve seen. And I’ve created an Upwork proposal skill like this one. Write a high converting Upwork proposal for a job post the user provides. So let’s write / proposal inside of our AI workspace. And again, you’re also going to have access to this skill. It’s inside of the old skills folder. We’re then going to take this post right here that was submitted 3 hours ago. I’m going to Ctrl A everything so we can see what it looks like. Going to paste this in and then say write a proposal, please. This job might not be the best one. You can see right here it says $10 as a fixed price, but I think what they’re trying to do right now with this post is literally just to find someone that they could maybe hire. You can see they’ve spent in total 9.3K. This is also what our skill flags. You can see $10 fixed price. Client’s average hourly paid is $7 and they have 67 open jobs. They’re probably just using this for a hire. That’s actually a good flag by claw. So, this is probably not the best job to spend your connects on. You can also see what country these people are making the postal from. Nigeria right here, which might not be the best type of client. You can see this one might be interesting. This is high level automation fulfillment partner. So, this is more of like a long-term job. You can see they’ve spent $50,000. It’s inside of go high level as well. So, let’s apply to this one instead. Maybe we can click apply now. Then we can copy this entire thing. I’ll write clear here to clear it. And then choose the Upwork proposal generator again. Paste this in and say build a proposal for this. And now it’ll ask for some information about yourself since it doesn’t have any saved facts. Have you built highle voice agents before? Yes, high level. I know everything about go high level and I have built a lot of voice agents in 11 labs not only in English but also in many other languages.

Do you actually use high level and it ends today? Yes, all the time. Pricing approach. Do you want to quote each service line by line? Blah blah blah. On hourly, one hourly rate is fine.

Your unfair advantage angle for fulfillment partner. The US time zone overlap pri fulfillment experience in-house de custom calculators. Sign off name Albert. Yes. Your unfair advantage. Let’s come up with an unfair advantage. prior agency fulfillment used to run my own agency. There we go. We give Claude some context and it’s going to build out this entire proposal for us. We don’t want to paste this in one to one. I’m going to copy this right here. You can see this would cost 27 connects schedule a rate increase. Let’s do never for that. You can always tell that later if you want. And then a quick tip is that you can write this is not written by AI. Then I don’t think this is good. will just write I used to run my own automation agency so the usell I fulfill model the fulfill is not that good so I’ll just write so I so I know the usell I fulfill model and when it comes to these proposals you want to educate them so I’ll write it’s important that we do good onboarding flow snapshot deployment sub account hygiene flows and snapshot deployments and then add some information that they might not know already we can actually use the go high level API through something like make it in to auto deploy sub accounts in both go high level and 11 labs. How start? Let’s first order your current highle setup. Then we built the reusable template that also deploys through Nad and go high level and 11labs through the API and then take over all let’s do like this this not like what I would write take all then I take over all clients world plus two close deals rate $16 an hour of quote fixed per client when I see The current onboarding process maybe just like current onboarding. This is a pretty bad question. There’s also some weird spacing right here. Let’s remove that too. One question. Is it custom build per project? Is it custom or do you sell productized

services will help me to plan it out? Albert like this right here. Then we can add an attachment like we can for for example add that calendar right here that is completely full. And there we go. This is a really good upwork proposal. What really can make it better is that if we also at the end right here, right, made a loom as well that describes how we can set this up. And then in this loom, you want to showcase that you actually know what it is that you’re talking about. Let me let me show you an example of what loom we can create for this. For example, I’m going to go to loom.com and then I’m also going to go to make.com, which is probably what I’ll be using to set this up. And again, this is becomes very easy when you already know what it is that you are doing. And again, in these proposals, you kind of want to educate them on what is actually possible for them to do. So, what I’m going to do now is that I’m going to make a loom video right here. Very important that we show our face. So, we can click record a loom. We turn on our camera right here. There we go. Make us as big as possible. That’s a good idea. So, they see, okay, we are an actual person, not an AI. There we go. Make sure to make yourself big in the corner. And then click start recording. Tire screen. Sup, guys? Just wanted to elaborate a bit about what I explained in the proposal. What you want to do is probably use either make.com or inn. It’s possible in both, but many don’t know that inside of go high level. And this is actually make specific. This doesn’t exist in inn. You have this API call right here that says adds an account. And what this basically does is that this adds a sub account automatically. So you don’t have to create one inside of um inside of go high level. And then what you can also do after you’ve used this is that you can use the go highle v2 API this one right here. And you have a an account update right here. So you can basically do anything you can do manually inside of go high level. You can do this through the highle API not something that’s inside of actual make. You would need to create an https request like this uh make a request but then you can basically do all of these things. You can update every single thing inside of high level. You can connect with uh other services as well. We can basically automate the entire thing. So, uh so yeah, just want to make sure that you guys know that this exists. And uh yeah, let me know if you need help in setting it up. There we go. That was what a 1 minute 15 second loom. Just copy this link right here and then paste it at the bottom. This is a perfect proposal. This is not written by AI. I used to run my own AI automation agency. So the usell model. So I actually really like the UCL model. It’s important that we do good on flows and snapshot deployments. We can actually use the go high level API through something like M.N. to also deploy sub accounts in both go high level and level apps. How start? Let’s first order your current high level setup. Then we built a reusable template. It’s sometimes good as well to write that we have a loom down here at the bottom. This not written by AI made a loom below that explains but I used to run like this right here. You want to make it seem like it’s not written by AI and all of a sudden we have changed a lot about it ourselves. But this is a perfect proposal. We have added an image of a screenshot of a book calendar. We don’t want to bid contacts. And then we click send for 27 connects. There we go. Submit. We can even see if it has been open yet. So far there’s 19 proposals, six open, 13 unopened. And all proposals are at $25 an hour. The top rated one is also at $25 an hour. So, we actually have a real shot of landing this job. That’s just what you want to do. You want to apply to a bunch of jobs and you want to stack up on connects. And you might think, well, I’m going to have to keep buying connects all the time. And that is honestly true. But there’s no other platform really like Upwork where you can find so many interested hot leads. And honestly, it’s very, very cheap. If you had to run Facebook ads, for example, for these type of leads, you’ll probably paid $100 or $200 per lead. Right now, for 10 connects, we’re paying $1.50. So, make this your routine. go out, send a bunch of requests, actually make them good, spend some time on them, spend at least like 10 to 15 minutes on proposal. And of course, at least when you’re starting out, and then when you get better, you of course going to be able to do it faster. Make a Loom video to every single one. This will help you land jobs. And you might think, Albert, well, already to this there was 20 applicants already. No one is going to choose me. Let me tell you why that’s not the case. And the best way to show you that is actually not to hop on my freelancer profile right here, but to hop on my buyer profile because I’ve also spent thousands of dollars on freelancers on Upwork. So I know what I like to look for when it comes to freelancers. Let me show you that perspective as well, so you actually know what your profile looks like and what buyers think about when they look for freelancers. So right here is now from my other profile. So from my company profile, not from the freelancer account, but from the company profile. And I made a post a couple of weeks ago about finding a video editor to basically cut up videos. And you can see I got 11 proposals on this post. And you can see who actually signed up. One from Pakistan, another one from Pakistan, a third one, fourth, fifth, sixth, seventh, one from Nigeria, one from India, one from Pakistan, another one from Pakistan. And the reason that your competition is not as good as you think is that a lot of the freelancers already on Upwork are from third world countries. Which means that if you are from western countries, you definitely have an advantage. This doesn’t mean that if you’re from India that you cannot do it. You’re just going to have harder competition because there’s a lot more freelancers from those countries already on the platform. And let’s take a look at the actual proposals. You saw how I reworked mine to not sound like AI. But what you’ll find is that almost everyone right here just has basic proposals as a top rated plus video editor with over six years of retention retention focused end to end YouTube postp production like it just sounds AI made then he has inserted a bunch of videos that doesn’t really concern us in terms of AI it’s a core part of my workflows I use tools like JBT I’m working in EST we can discuss the monthly rate that’s it I love to connect blah blah blah blah you can see very AI generated not very personal this one as well portfolio at the end of the process like this is just like a template. You can see that it’s a template. It’s not personalized to the actual job. That’s just to show you from the buyer side as well how it actually looks. So you know, okay, this is what I should write. This is how it should catch their attention. The most important part of your cover letter as well is this part right here that we wrote, right? This cover letter because this is the thing that they see and determine, okay, am I actually interested in this person? And it also shows how important it is to actually just land your first couple of jobs on Upwork. And the thing is, when you’re just starting on Upwork, you’re going to have to do more effort, right? You’re going to have to make a better cover letter. You’re going to have to make it more personalized, send more looms. But what you can see right here is that this is definitely what moves the needle like the job success rate, how many jobs completed, total hours, and also money earned. The bias on Upwork, they definitely look for that. And then to set some expectations, building an Upwork profile can take months, if not years. It’s not a get-rich quick thing by no means. It also takes long work. But what you’ll find out is that these clients that you work with on Upwork, all of a sudden they refer you to someone else and someone else and someone else. And Upwork, even though it is a freelancer platform, it’s the perfect place to start. Awesome. I showed you this just to make you understand that the competition on Upwork isn’t really that bad. And Upwork, even though that it’s a freelancing platform, is in my opinion one of the best places to start. And here’s why. You build experience. It’s what I mentioned in the start, right? Get good before you get rich. And this by actually doing the work is the best way of getting good. And what also happens is that all of a sudden you have clients, you talk to people and you will be referred to other clients. It’s a snowballing effect. And all of a sudden when you have a bunch of clients, you can automate a lot of the work with cloud code. And boom, you’re running a one-person business with clawed code. So if you want to do any type of outreach, I recommend that you start with Upwork. And just to set some expectations, it might take a couple of weeks before you land your first freelancing client, but it’s still going to be so worth it. However, if you want to speed that up and get even more volume, then I’m going to show you something insane that we can do with Cloud Code in the next part of the course. Let’s get into it. Great. To be clear, now you just want to send as many connections as you want and land your first couple of jobs on Upwork. That’s why you have the highest likelihood of getting your first client the fastest. If you want to push that further and send out a bunch of connections to every new job that is posted on Upwork that you’re interested in and you want to do even more than that, then this part of the course is for you. If not, that’s completely fine and you can skip to the next part of the course. Do you remember when we talked about this map right here where all of these countries marked with green, they are getting very very used to AI, especially like chat bots and voice agents. All the other countries however are a bit behind. So that is South America like for example South Africa is a very booming market. It’s especially Europe and then also many parts of Asia. If you live in one of the blue countries then you should especially listen up in the next part of the course. This will also work in the green countries but it will have lower reply rates. What I’ve seen is that this next method is absolutely insane in the blue countries right here. And what we’ll be doing is something called cold email. You have probably heard about it, which is where we can send a lot a lot of cold emails out to a lot of qualified prospects and then the people that are interested will reply back and we can automate this entire thing. My favorite code emailing software is instantly.ai. I’m not sponsored to say this. That is just my favorite platform. If you want to set this up and you want to support this channel, then make sure to check our affiliate link right below this video. Again, I’m not sponsored, but I’ve just created an affiliate link through their affiliate platform. But why is this software so good? Well, it allows us to get a bunch of email accounts very very easily and then send a bunch of bunch of emails to qualified leads and then they have an MCP that allows Claude to do 90% of the work. So, let’s set that up first, shall we? The first thing I want you to do is actually search for the Claude desktop app if you don’t have it already and then download it for whatever computer you have. So, I’m going to search Mac and then download Claude right here. Download for Mac OS. That’s going to download the installer just like this. The next thing we want to do is actually not go inside of Claude code, but then go inside of the Claude desktop app that we have just installed. The reason we do this is because that’s by far the easiest way to add MCPS. If you go in the bottom left corner right here and then click on settings and then go inside of connectors and then click this customize right here cuz they have now moved it. We can see all of the connectors that we have. So I can now click add connector right here. Add custom connector. call this instantly which is the platform and then what we need to do now is to go back to instantly. So now’s the time if you haven’t already then sign up for an instantly account again. If you want to use my affiliate link then that would be very appreciated. We can now click on the bottom left corner right here go to settings then click on integrations and then click on API keys right here. And now we want to create an API key. This is what allows claude to basically have access into instantly. So click create API key. I’m going to call this claude. I’m going to select all so I can do everything it needs through the API. And then we click create. And this is going to create this API key right here. So let me copy that one. And then I’m going to go to this docs right here. I’m going to leave this docs right below this video. But that’s basically the MCP URL that we are using for instantly. So you can see it says your API key right here. You will copy this, change it with your own API key. And of course I’m going to rotate this API key after we’re done. I can now copy this entire thing. Go back to plot, add it right here, and then click add. And there we go. The instantly MCB has now been added. I can now just go back if I wanted to test it right quick. Go into cloud code, for example, and just ask it right quick. Do you have access to the instantly MCP? Now, this is just to verify that it actually has the access that it needs. You can see it says yes, the instant MCP is available. And I can see tools like blah blah blah blah. There we go. I’m also just going to ask it right now, can you pull campaigns just to check the connection works. This is to ensure that it both has the MCP. We just added that, but that it also can list all of the campaigns. You can see it’s using this right here. List campaigns. Let’s see if it returns. Connection works. One campaign returns blah blah blah like this. Great. The reason we installed the MCP inside of the claw app is because that makes it much much easier when we go to claw code. I’m going to click Ctrl C right here a couple of times. That’s going to close down cloud. I’m going to write clear again and write cloud again. And that’s basically how we restart cloud. Now I can write slashmcp. Hit enter. And under cloud AI, we can now see the MCPS that we have added. And you can see right here cloud.AI instantly is connected with 38 tools. I can hit enter on that as well. I can hit enter to view tools. And then we can see all of the tools that it has access to which was what 38 tools, right? They can basically reply to emails, list emails, create lead list. They can do it all. And you can see the status is already connected because we added the API key right here. Cool. We can write clear. That clears the claw code session. And then we can basically get FL to create these cold email campaigns for us. Before we do that, we need to do a very important step, which is we need to go into instantly and get some email accounts. Email accounts are basically the emails that we are sending from. And there’s two different ways that you can do this. I can click add new right here. You can either choose pre-warmed accounts, which is basically accounts that instantly has warmed up for us. These accounts don’t have our own name. You can see it’s Arthur, it’s Charlotte, and it’s using these interesting domains right here. But they are pre-warmed, which means that when we send emails, it will have high deliverability. And those are the ones that I recommend that you start with because usually you would have to warm up your own domains, which takes a lot of work in 30 days. Here we can just pre- buy them. You could also connect existing accounts if you want to, but for now we just going to choose the pre-warmed accounts. We can click go. Now we can select some domains. We want to find something that seems very uh professional like land aiimag.org. That’s probably pretty good for an AI agency, right? I’m just going to choose landi imagine. It’s $15 right here. And here you can see we get five emails. You will set the forwarding domain. And this is very important that you set it to your own website because if people then go to this website, it will go to your own. So I’m going to go to my own website, shiny.ai. AI and to paste that in right here where it says enter valid URL and then click place order. And now we just need to check out right here. You can see it’s $65 to get these domains, but that is very very worth it. It’s $10 per domain. When you have bought your email domains, it’s going to look like this. Now, you can see we have five email domains right here from this domain right here, frivegateway.com. And we can see the health score right here next to them. You can see 100% health score. Each of these, you can see it says email sent can send up to 20 emails a day. So when we have five emails, then we can send 100 emails a day, which will be plenty because we’re going to take a more targeted approach with our cold email. We want to get this right. So after you have your domains and after they are at 100% health score, we’re ready to start and create our campaigns. And of course, we’re going to be using Claude code for that. And I’ve actually already created a skill that’s based on ours most high performing campaigns that you of course also have access to. So let’s get started with that. Before I show you how to set it up with Claude, I wanted to show you this campaign right here, which is what you should strive towards. And this is also what the Claude skill is based on. You can see this campaign right here is going out to dentists. And we are 60% through. We have sent a,000 emails. This is both initial emails and then follow-ups. Follow-ups in cold email is very important. And from that, we have 50 people that replied with a 5.1% reply rate. 5.1 is extremely high in cold email. Usually when you send code email campaigns, you maybe get a 0.5% reply rate with a lot of people that say [ __ ] you or I don’t want to hear from you. But from those 50 replies, a lot of them will also take me off the list or don’t talk to me again. That is just a part of cold email. But from those 50, we have six interested leads. These are people that said yes, I’m interested. Let’s hop on a call. Let’s talk more about this or asked a follow-up questions that we can then answer and start a conversation with. And you can see if you can send let’s say a 100 emails a day, you get five that replies that means that every other day you get an interested lead which is extremely good and you can run a business on that. So let’s get into it. Let me show you how we create this campaign. Awesome. If we open a new Claude code chat, write instantly campaign like this. This is the instantly campaign creation skill. We can then write I want to create a campaign for let’s do Rufus in Sweden. And if you are from one of these blue countries that I showed earlier, then usually what a lot of beginner do is that they still go after the English markets. So after the US, the UK, Australia, but because AI is being adopted so massively everywhere around the world, I would actually say that it’s a much better advantage that you speak some unique language. For example, if you speak Swedish, please go after Sweden. Just make sure that it’s legal to reach out with cold email in the country that you want to go after. But please go after those countries. Why? Well, you’re going to have a much higher reply rate. For example, I don’t really speak Swedish. I understand a little bit, but someone that speaks Swedish have an unfair advantage over me by knowing the language. Same if you’re from Greece or from Portugal or from Belgium, then you speak a language that not the entire world speaks, which basically just means that you have way less competition. So, just to do it as an example and to show you that Claude is going to do all of the creation of a campaign for us, I’m going to create a campaign for roofers in Sweden while not speaking Swedish, only understanding a little bit. And that’s all we need to write. Now, we hit enter and Outlot code is going to build out the campaign using the instantly MCP that we have connected it to. And you can see it says, “What are you selling to Swedish roofers? Describe the offer in one paragraph. What’s what does it do for them?” Let’s do new website just to show you how it looks. What’s the primary pain? Slow quoting loses jobs, few Google reviews. Let’s do few Let’s do a couple like few Google reviews, ranking on SEO, and also missing out on leads because they don’t have the right information. What tangible asset will email free offer? This becomes the main subject line. Let’s do the lost revenue calculator so they can see, okay, how much they’re saving. What should the campaign be called? This doesn’t really matter, but let’s just call it roofer Sweden. And then submit answers. And now P code is going to build out this entire campaign for us. And here we go. We have a draft that goes over everything. This means I was thinking about calling, but I wanted to write to you first. And then saying right here that I’m thinking that you’re losing jobs because your website is not showing high on Google. Is this something that you would need some help for? I really like this. This all absolutely awesome. So I’m going to approve. Let’s create campaign. And now plot is going to build out the instantly campaign for us. We don’t have to do a thing. Now you can see it is calling instantly. If I click control O, we can see what it does. It’s making this request right here with the information. There we go. Campaign is created and called Rufus Sweden. If we go into instantly right now, hit to campaigns. We can see that the new campaign has now been created. If we click in on this, we can go to sequences, which is basically where we write the emails. But it looks completely blank for some reason. It looks completely blank. I don’t see the emails. Let’s see what happened here. Oh, I just had to refresh it. Do a hard refresh. Shift command R. That half refreshes the page. Now, we have the entire campaign inside of instantly that we just wrote out. So, we need to make sure that we click save. There we go. Click save on everyone. Now, we have our campaign ready to go. The last thing we need is just leads to send to. and I’m going to show you the fastest and what I found to be one of the cheapest way to get really qualified and verified leads. Let’s get into it. To get qualified leads, we’re going to use a platform called Apollo. But don’t worry because Apollo itself is actually extremely expensive. I’ll show you a cheaper way to get the same quality leads as Apollo, but using a different site. So, let’s say we want to find roofing and building companies in Sweden, for example. We then make sure to go to people right here. We go to location and we search Sweden. Then we go to industry and keywords and we could search for that will probably be construction. Let’s maybe also do real estate. If it’s like home builders, then real estate is probably also going to fit. You can see now we have a 102,000 people. You probably want to narrow it further. So you can also do things like job titles. you of course want to do go for someone like the owner of the business or the founder of the business. Add a lot of keywords to people that would fit your description for your ideal client like CEO, founder, CEO and founder. Add all of these because you can see that increases the list size right here. You could also exclude titles if you want to. But now you can see we have a list of around 3,000 people right now. What’s very important that you go into email status and then make sure to click this verified right here. That’s going to get you the best quality of emails. And you can see our list is now at 2.1,000 people. You can also choose if it should be B2B only, B2C only, if it should be e e-commerce companies, fintech companies, nonprofits, SAS, consulting. You have a bunch of things to choose from inside of here. This one is also one of the most important filters. You can choose what range you want to reach out to. So, how big the company should be. If you’re just starting out, you’re probably not going to close an enterprise deal right off the bat. That would be insane. We could probably choose like from 1 to 10 to 11 to 20, maybe up to 21 to 50. And that gives us 1.8,000 people that we could reach out to. That way, we leave out the biggest companies because you can reach out to those in the future, but you need some experience first. And there we go. Now, we have a bunch of Swedish companies and we can access all of their emails. But usually, if you’re using Apollo, you do it directly through the Apollo platform, which is extremely expensive. What you can do instead is that you can use this platform right here that’s called trustedleads.io enterprisegrade B2B lead list starting at 0.005 per lead. So go in there, create a user. Again, I’m not affiliated to say this and then you can click new order right here. Here you write in your name, your email address, write in a phone number and then you want to choose the Apollo URL right here and the number of leads that you want to scrape. Very important that you turn on the verified leads that I showed you earlier. Otherwise, you’re going to get bad lead quality, and you’re going to be paying for leads that are not good enough. So, let’s say a total of how many do we want to scrape? Like 1,500 of these leads. And you can see that’s going to cost us $25, which is 67% less of what we would use on Apollo. So, now we can click continue. Now, we want to take this URL from Apollo right here and paste it in. It says that the email is not verified, but I’m almost 100% sure that we set it to verified. You can try and refresh this page just to make sure that you have the newest URL like this. Let’s copy this one. Paste this in and see if it says now. That is fine. Let’s just verify that we have the right search. We do. Then click continue. You could if you wanted to pay additional to actually verify the emails, but I’m going to show you a better way of doing that. So, we’re going to click no additional verification. Do a signature and then click proceed to payment. After you’re done with that, then it can take a couple of hours, but then you receive an email like this. Your order is ready. Thank you for your order of B2B leads. download the lead lead list right here. And this is a previous list that I bought and this is actually 10,000 leads. You can see how many leads we have right here. When you buy leads from inside of here, it is very smart to do it in bulk. You can see here we have a lead list of 10,000 leads. What I’m going to do now is that I want this as a CSV file. So I can click file. I can click import. Click upload. I can upload this CSV file right here. That’s going to take some time cuz it’s a very big file. Replace spreadsheet and import the data. And here we go. Here we have our big list. We have to do one more step. For now, I don’t need all of the rest of these leads. So, I’m just going to delete those like this. This is just to show you how you import those into instantly. I’m going to call them leads and maybe just the date May 16th. And I’m going to click file, click download as CSV. And then before we put those into instantly, we want to run them through million verifier first. This is basically finding all the emails that are old. When you’re sending out cold email campaigns, you don’t want to send to email domains that doesn’t exist. That’s like a tell that you are spamming with emails which the email providers don’t like. Instead, you want to run it through something like million verifier. First, you click select file and we can select this one right here. And you can see we’re going to be using 969 credits for this. And we’re going to be removing the duplicates. We click start verify. And this will now create a job right here. You can see it says takes 15 minutes. And this is basically going to find all of the duplicates and it’s going to remove all of the old emails. It’s basically verifying the emails to make sure that they are good to send to. You don’t want to skip this step. That’s very important because if you skip it, then you’re going to be sending emails and your domains will instantly be backlisted and then you’re going to have to buy new domains. So, make sure to use Million Verify as well. Again, you can find links to all of these platforms in the description of this video. There we go. That took around 10 minutes. And you can see now we have a total of 64% of emails that were good emails, 17% that are risky, and 18% that are bad emails that doesn’t exist anymore. You definitely don’t want to be sending to these bad emails cuz those emails are going to bounce and it’s going to hurt your domains. Risky emails you can sometimes send to, but I would still not recommend sending to these inboxes. And then we have the 64% that are good emails. Those are the ones that we’ll be sending to. I can now click download report right here. And I want the good emails only. So I’m going to click right here. That’s going to give me this file right here of only the good emails. There we have it. And now it’s time to add these emails to instantly. I’m going to go back into instantly. I’m going to click on the leads tab under my campaign. I’m going to click add leads and then from a CSV file and then I’m going to choose the CSV file that we just downloaded which is this one right here from millinverify.com. And you can see we now have to map the fields with first name is going to be first name, last name, last name. Title is going to be job title, LinkedIn URL, email, all of these we don’t want to import. Let’s see if there’s any one we need. Probably not. Company name we want. website. We want industry probably not. And we don’t want to verify our leads because we just did that ourselves. Now we can click upload all. This will upload 625 contacts to your campaign. Let’s click yes. And now you can see all of our leads are uploaded to this campaign right here. These emails I got done probably a couple of months ago. So we of course not going to send emails in Swedish to these US leads. And then when we have our leads, last thing we need to do is go inside of options and then just add our email accounts like this. You can see these email accounts right now are inactive, but when you’ve just gotten them, they’re going to be good to go. And then you can click launch and publish like this. When you have your sequence in place, when you have your leads uploaded, when you have your pre-warmed email accounts, then you can click start campaign up here in the right corner, and that’s going to send out all of these emails. Another thing that you can change if you want to, you can see this is set from 7:00 a.m. to 400 p.m. by default and the time zone right here. This is a pretty good window, but if you wanted to change that, you could of course do that. You can also send on weekends if you want to, but I probably wouldn’t recommend that. Do it in business hours. And here’s what’s going to happen. When you start sending out, you’re going to start receiving replies. And if you’re sending to these countries that only you speak the where you speak the language and not that many else do, then you’re going to get a very high reply rate. If you have made a good campaign with a good offer, you’ll find those responses inside of Unibox. You can click on interested and then you can see the responses that are interested. And you can see this is from a real campaign that we sent around a month ago. And what you can see is that we are receiving interested responses. This says yes, I’m interested in a conversation. This person right here says that they’re interested in an offer to see what we provide. And in this campaign, we have only sent to 278 leads. So, I hope you start to see the power of cob building, especially to countries where you speak a unique language. If you want to absolutely maximize so you get as many clients as possible, I recommend running Upwork as your main priority every single day. Go out and send connection requests because those are the hottest leads. And the easiest way to gain that initial experience, which is so crucial. And then if you want to maximize, then also run a campaign in the background cuz this is just going to be running sending out emails as you go and probably book you a couple of meetings a day if you have a good offer and offering an interesting service in your market. These two things when you’re just starting out is the highest leverage thing that you can do. But I honestly recommend getting started on that right now. But you of course also need to know how you can actually convert meetings and how you take this information that you’re getting on meetings when people are actually interested in proceeding. how you take all of this information and give that to Claude so Claude can build out and do the majority of our service delivery. This is going to be extremely valuable for you to watch. So let’s get into it. Great. Now you have outreach going out on both Upwork and on instantly. The next part of this course is one of the most important because this is where most beginners they mess up and they mess up in three places. Both before the call, under the call, and after the call. So, let’s go through some ground rules that you need to understand before you start booking meetings and closing clients. The first one is that you never, and I mean it, never mention price before the call. And here’s the reason why you don’t do that. If you mention the price, let’s say that the client asks for, “Okay, what is the price before I hop on a call?” And you say, “Well, it’s $2,000 a month.” What then happens is that the client judges you on the price and not the value. The client doesn’t know enough about your service. They don’t know what work you’ll be doing and why it’ll give them a bigger return than the price that they’ll pay. So, when they hear a $2,000 price, what they think about is, “Okay, do I want to pay $2,000 right now?” Probably not. I’m going to stop replying to this guy. And the only reason they do that is because they don’t know your value yet. You only mention the price after you’re on a call with them, after they know how you can help them. because it’s a way better equation, a way better thought process if you show, let’s say, $10,000 worth of value and then show the $2,000 price point. All of a sudden, that becomes a no-brainer deal. So many beginners make this mistake and they throw out a bunch of meetings because they don’t understand this key principle. That is rule one. Rule two, that you always want to send a confirmation and reminders. It happens more often that you think that people simply just forgets that they book a meeting with you and then they get kind of like embarrassed that they forgot the meeting and you never hear from them again. You always want to send both a confirmation and reminders to every single call that you hop on. So the process looks like this. You have people interested either over cold email or over Upwork. You decide to okay, let’s hop on a meeting. And then as soon as they have agreed to a time, you send them the confirmation. Right? This is 400 p.m. Monday. This is where we meet. And you send this on as many channels as possible. So if you have them on email and their phone number, then you send it on both as many channels as you can. You do this instantly. And then depending how long the meeting is out in the future, you then also send reminders. I recommend that you send reminders at least every 3 days up to the meeting. Then do it one day before and then do it 1 hour before and then do it 5 minutes before. And the way that these messages look is basically just like, “Hey, first name, just reminding you that we have our call tomorrow. Excited to see you there.” If you’re doing like a demo for them already, it’s a very good idea to mention that you’ve also put worked in before the meeting because that then makes them feel guilty if they don’t show up. So, if you build a demo or a test website for them or something ahead of time, then say, “I’ve prepared a demo for you based on your own company.” So, you can see exactly how this will look for your business. If they then don’t show up, they’re going to feel guilty. You want to include that in the reminders. And then the last reminder, like 5 minutes before, you should just write, “Hey, first name. I’m just finishing up another call. I’ll see you in our call in like 5 minutes.” That’s a really good message for this last reminder. Something I also see, and this is connected to never mention price, is that some people want to skip this step of hopping on a call. The Reddit is that’s just making your life 10 times harder. Not everyone’s going to be ready to buy right in that moment. They might need more information. So, you need to hop on a call with them. You want to give them limited amount of information. So where they really get the information is on the call with you. And it’s very simple. If you have an issue right now where people don’t show up to the calls, it’s just an equation of do they think that this call would be valuable enough for their time. Do they think it can actually help them? And if you have an issue where people don’t show up to the calls, the reality is they don’t think that it’s going to be worth their time. That is why you need to include the reminders. You need to tell them, I’ve built a demo specifically for your business. I’ve prepared this and this and this. We’re going to see this and this and this. And you of course need to solve a painful problem. Okay, so before the call, you never mention the price. You always want to be judged on value. You always send confirmations and reminders on as many platforms as you possibly can. And then you always never skip the call. You don’t sell straight away. You hop on a call with them before you start selling. Okay, now on the call. A lot of beginners make the mistake of talking too much. The best way I found to sell these AI services is not by being the one talking. is by taking what I call the doctor approach. When you go to the doctor, they don’t just hand out medicine the moment you come in the door. The first thing they do is that they ask you a bunch of questions like, “Where does it hurt? How long has it been hurting?” And then they try to diagnose what’s wrong. You want to take the same approach when you are selling. And the way you do that is that you only talk 20% of the time and they should talk 80% of the time. And when you talk, it should mainly be questions. And when they talk, they should answer your questions. And here’s why that works so well. When you’re the one asking the questions, you’re the one leading the conversation. And one thing is that you mention why your service is valuable. An even better thing is when the client themselves mention why your service is valuable. When you get people to say themselves that they need something, they convince themselves way better. And the way to do that is by asking questions. Let me give you an example. If someone has, let’s say, only a couple of reviews on Google Maps, you can then ask a question like, “Okay, what do you think happen when a customer compares you to your competitor? Your competitor has 150 reviews. I can see it right here. You only have five reviews and one of them is a freestyle review. Who do you think your client is going to go with?” They’re going to say, “Well, based on this only, they’re probably going to go with that client.” Do you see what we do there? We get them to mention themselves that they have a problem. And when we take this doctor approach, it becomes a much stronger cell. And it doesn’t even feel like selling. You’re just saying, well, what do you think happens when when this and this and this in your current situation, they mention a problem. And then it makes it much easier later in the call to say, well, I actually have a solution to your problem. That’s the entire point of the call. It’s to position a problem that they actually do have. If someone doesn’t have a lot of reviews on Google, they have a problem that’s hurting their business. And then position you as the solution. That way, you don’t need to convince them. You don’t need to be overly salesy or talk a bunch. You’re literally just asking good questions. And by asking good questions, you control the conversation. You can steer it in any way that you want. And that leads me to the second rule, which is that you need some structure on your calls. And the structure that I like to use is initially just these doctor questions. Initially, just asking if they even have a problem. And what you’ll run into sometimes, let’s say that you are providing more Google reviews with automation and you run into someone that already has 500 reviews and are on top of everyone else. They probably don’t need that specific service. You might sell them another service, but you don’t want to sell something to someone that doesn’t actually need your service. That’s unethical. So, the structure that I like to use is one, ask a bunch of good questions that positions a pain. Then two, tell them that well, we can take this pain and we might have a solution. Let’s see if it even makes sense. And then you can show a demo or you can explain well we can actually do this to fix this problem. And three is that you then explain further about this solution and say well the price of this solution is and only when they’ve seen the solution when they have seen the value that you can provide that’s when you mention the price and that’s when you shut up. They now get themselves to decide okay this pain that I have and this solution right here is this worth it or is it not worth it? And what will usually happen is that they will then have some kind of objections and those come at the end. Objections are usually like questions. Well, what happens if we try this for a month and it doesn’t work? Or well, what happens with my current website if we’re changing it over? How would I do that? It’s basically just questions from their side that show some kind of concern. So, what you then have to do is just answer those questions in a good way and then make sure that that’s that concern is squashed and then at the end that’s when you close. That is the structure that I like to use where it doesn’t really feel like a sales call at all. This is what you should do. Let me show you what you then shouldn’t do. A lot of beginners they make the mistake of basically just hopping a call and just yapping. So a lot of beginners structure look like this. They just explain their product and usually they basically just explain features and then they ask do you want to buy after they’ve done that. Right? So what does this look like? This is maybe if you create a slideshow that’s just about your product. this is what we do and you just sit there and and go through it. The problem with this approach is that you don’t position a pain. You don’t ask good questions. You don’t make sure that they actually need your service. You basically let them make the decision themselves if okay, do I need this or do I not need this? And if you have a boring slideshow, chances are that they’re just going to zone out. All of a sudden, they hear a $2,000 price point and they say no. All right? So, don’t do this. Do this structure instead. Right? Questions first that position a pain. This is questions like, well, your competitor has way more Google reviews than you. They have a good website as well. You have almost no reviews. You have a trash website. What do you think happens? Well, don’t say trash website. Say, your website is quite old. Who do you think your client is going to go with? And just like a first impression. You want to say themselves, well, well, in this scenario, they’re probably going to go with my competitor. You position a pain that hurts their business. Then you position your solution. Well, we can install this AI that reaches out after every single job and gets you Google reviews. And we can also install this website. I can build it in 2 days, which looks way better than what you currently have. Do you think that would help increase your first impressions with new customers? Again, you even position that as a question to make sure that they understand the solution is worth the price. Great. Great that you now understand there’s so many beginners mess up right here. You don’t want to do that, but you now understand this, which is good. Great. So, now you’re starting to get sales. What do you do? Well, you follow rule number three, which is always book next call on this call. Always book next call on this call. Here’s the reason. When you’re just starting out and you don’t do this, there’s going to be a bunch of times where someone agrees, “Yeah, let’s go. Let’s do it.” And then you don’t book the next call on this call. And what then happens is that they ghost you. If they have something in their calendar, it increases the chances that they actually hop on a call with you. Again, this rule is extremely important and it’s going to save you a bunch of deals. So, what you basically just do on the call is that you say, “Great, happy that you are on board. Let’s book it into calendar already the next time we’re going to talk and then I’ll have this and this ready for you.” Again, you kind of sell the next call. You hype up the next call and you make sure that they see that you’ve made an investment now in them, that you’ve spent time so they feel bad if they don’t show up. And the last rule, and this honestly surprises me that people still do this to this day, is that they don’t take payment on the call. You always want to have some initial commitment. And even if you’re doing like a free trial, you still take the payment information for after just like you would with a software. If you get a free trial, you still give your payment information first. Do that here as well. The reason for that is that they need to make a commitment and they need to make a commitment on that call. And you can even write it into the contract that if they then know your next call that you then charge a $500 fee for wasting your time. Like you can write that into a contract, but you always want their payment information on file. And the way you would usually do that is with Stripe. And what we usually do is that we just get like a $1 payment link inside of Stripe. If you create a $1 a year payment link, what that does is that it saves the payment information. You can just say that’s like a $1 down. We’ll refund it to you. Don’t worry. But then you have that payment method on file, which means that you can charge it in time. And that also removes friction later on because now you have that payment method on file. So let’s say that you’re doing a commission deal or you’re just charging $2,000 a month. You can just use this payment information to set it up. they don’t have to do anything on their side, which again removes friction. So, you always want to make sure to take some kind of commitment on the call. Even if you’re doing a free trial, get the Stripe $1 payment link. Let me actually show you how to do that. You go inside of your Stripe dashboard. Then you click on payment links right here and click create payment link. Then you create a new product. You just call it checking card, something like that. You do it recurring and you do $1 and then you do a year like this. And then you click add product. You don’t need to collect tax automatically on this cuz you’re going to be refunding the payment anyways. Create link. And what will then happen is that when someone signs up on this link, you get this payment link right here, which says $1 a year. When someone fills out the information right here, you get their email and you also get their payment information. And then it allows you to later to just create a subscription for whatever you’re charging. Let’s say you’re charging $2,000 a month. You can then go in and set up that payment without involving them. So that’s how you create Stripe payment links. And trust me, that’s going to help you get commitment, which is going to help you close more deals. If you have a problem closing, you’re using this format right here and you get to the point where you mention the price and people don’t buy it, then use this last tip, which is kind of a bonus tip cuz it’s not always required. Sometimes you have such a good offer that this is not needed, but I call it the proof of concept offer. What is this? Well, this is an offer that has made us tens of thousands of dollars. Here’s how it works. It’s basically a satisfaction guarantee. And you can say this that well, this is our satisfaction guarantee. What you say is that you get payment up front. You always want to get commitment, right? So, let’s say that you are selling a website. You still get your payment or whatever you agree on. Let’s say it’s $1,000 first and $1,000 later. You still get your $1,000 upfront. And you take that payment on the call via Stripe, but then you give a satisfaction guarantee, which means that I’m going to build this entire thing out. So, if we use the website example, I’m going to build out the entire website for you. So, you can see everything that’s going to be on your website. You can see the finished product. If you’re not happy with your website at that point, then you’ll get a full refund. So, if you don’t even want to see how it performs when we give it to customers and when we actually launch it, you can get a full refund, no questions asked. We given this offer a bunch of times. And the great thing about this is that one, it removes all risk because they can actually make an informed decision. They can see exactly what they’re getting. And if they don’t like it, they can get a full refund. And two, it makes you seem very confident and good at what you do. All of a sudden they say, “Okay, this guy can put money on the line. He will refund me the entire thing if I don’t like what he creates.” That makes you seem very confident, which again is going to increase your close rate. We have given this offer so many times, and we have never had to refund anyone because people have been very happy with what we build. So, if you’re good at what you do, this is such a no-brainer offer, and adding this will probably just bump up your close rate with like 10% instantly. And you can use this for any AI build really. It doesn’t matter if you’re building websites, it doesn’t matter if you’re building automations, you just want to hop on a call with them and show it face to face. Okay, here’s how it works. Look at this. It does this, does this, and does this. Get them to try it. Get them to see how good it actually is. And then say, great, let’s launch it. You don’t even need to mention the satisfaction guarantee. Just mention it on the call when you close them. And they’re going to come back to you and say, well, if they don’t want it, then they can get it refunded. But as soon as they commit, and as soon as they see, okay, how good of a job that you have actually done, they’re not going to want the refund. They’re going to want to actually implement it. Great. So on the call, you only talk 20% of the time. They talk 80% of the time because you use the doctor approach. You use this structure right here where you start with questions. You position a pain. You then position that pain to your solution. Exactly. You handle any objections and you close the sale. You don’t want to just have a slideshow where you just explain your product because you don’t position a pain. They’re going to judge you on your price, not on your value, and they’re not going to feel like your solution is a fit to their exact problem. Then you always always book the next call on your current call. So, they need to know when they’re going to hear from you again. You don’t make the mistake of not taking payment on the call. You always take payment on the call. You get a payment information. You get the $1 Stripe link. And if you’re struggling with closing deals, then use the proof of concept offer because this removes all risk. You still get commitment up front. But if they don’t like the build afterwards, you basically just have a satisfaction guarantee where they can get a full refund. Awesome. After the call, then this shouldn’t come as a surprise, but you need to do a good job when you’re just starting out. your reputation and building trust and building up a client base is absolutely everything. Make sure that you do a hell of a job that you go all out for this client and do a hell of a job. Number two, a mistake that many beginners make as well is that they don’t continue the calls. Let’s say that you’ve set up an AI system. A lot of agencies, they just set it up and then they never communicate with the client again. That’s not a good way to do it. I’d recommend at least bi-weekly calls. So, hop on a call with them every two weeks. Go over performance. Go over things that could be improved. Why is that? Well, that’s going to decrease your turn a lot. Let’s say you’re charging $2,000 a month. Doing continuous calls by a weekly will probably decrease your turn to something like 5% or something. Where many agencies have upwards of 20 to 30% churn. If they have 30% churn, it means that a client on average only stays for 3 months. There’s another reason to why we do bi-weekly calls. This is two steps that almost every beginner’s day miss. They don’t upsell. Let’s say that you just helped someone build a very, very cool website. You don’t stop providing services there. Then you ask, well, do you also want an AI to pick up the phone calls from the people that call you now because now you have a good website? Or, well, I just helped you build this website. Do you also want to send additional traffic with Google ads to this site to get more customers? You always keep upselling. And if you don’t have it already, then come up with ideas that could help their business further. Upselling is a way to increase your lifetime value, also called LTV, which is the total amount of money that your client pays you over their lifetime as a client. Upselling is the best way to increase LCV without increasing the cost of acquiring a client. You’ve already acquired this client. It doesn’t cost you anything more just to hop on a call with them and say, “Well, do you also want this?” And then charge for that as well, of course. And it doesn’t stop there. And again, another reason to why we do bi-weekly calls. You also want to ask for referrals. Referrals is another way to get more customers without paying for anything. You’re not paying for additional code emails. You’re not paying for additional connections on Upwork. This is free clients if you just ask for referrals. So many beginners when they start out don’t ask for upsells and they don’t ask for referrals. But that’s also because that they don’t do it the right way. Here’s how you do it. You don’t say, “Do you know anyone that could use this as well?” Instead, you’re very specific. You want to mention the exact people that you want to work with. Like, do you know any plumbers in your area that could use this as well? Or do you know anyone in this industry or in this industry that could use this? Like, do you know any plumbers, electricians, people that own a landscaping company? You basically just name a list of potential clients that you want to work with. The reason that you do this is if you just ask, well, do you know anyone else that could use this service? They might come with someone, but usually they’re just going to say no because it’s harder for the human race to just come up with something from scratch. If you give them a specific list, like, do you know any plumbers, electricians, landscapers, do you have any other friends in business? If you already give their brain that to think about, usually they’re going to come up with more referrals to you. That’s rule number one when asking for referrals. Rule number two is that you always want to give a commission. If someone refers you to a very very big company, you can give them 20% of your revenue. That is fine. You’re going to get way more referrals if you actually incentivize people to refer you, saying, “Well, you’re going to get 20% of whatever they pay me. I’m going to pay that to you.” Do that to all your clients and you’re going to get way more referrals. I hope you see how all of these are connected. You do a hell of a good job. You need to make the client happy. And a part of making the client happy and treating them well is also to giving them the time every two weeks. and only if you do a good job and they also want to refer you to their friends. It all stems from actually being good at what you do and doing a good job. Great. I hope you found this section of the course valuable. I would honestly take all of these things, write them down so you have them and so you remember them. Look over them daily cuz these rules are literally going to save you thousands of dollars if not tens of thousands of dollars through your AI agency career. All right, the next part of the course is also very important because that’s how you actually do the service delivery. And I have an insane trick that’s going to save you so much time prompting Claude and basically have the product delivery ready before you even leave the call with the client. It’s insane. Let’s get into that. What most agencies do when they are done after a call is that they get to work. And if you’re using something like Claude Code, you’re then taking the information that you yourself learned from that call and you’re passing that to Claude code, right? You’re spending time giving it all the context that you gained on the call. But human mistakes happen. You might have missed something. You might have forgotten something. Sometimes you don’t start on the project right after the call is finished, which means that you might lose out on some information. Let me show you both a smarter way and a way where you never miss information ever again. Where you pass all the context effectively to claw. To do this, we’re going to go back into the claw app that we installed earlier. If you don’t have it already, then install claw for desktop. You go to this settings tab right here. Then you go to connectors. And now they move it. So you go into customize. That’s going to show all your MCPS right here. Click this plus right here and click browse connectors. And then search for Fathom or the noteaker that you’re using. On my calls, I’m using something called that’s called Fathom which basically records my calls and it gives me a transcript. This makes it very very easy for us to just say, “Hey Lord, I just finished the call. Go in and check the transcript and get all of the context.” I’ll recommend that you click this from needs approval to always allow. And now the Fathom MCP is connected. So if we go down and restart Claude because every time you install a new MCP, you need to restart Claude. And we then write /MCP, you can now see that the Fathom MCP has been connected. And if I click in on it and I click view tools, you can see we can get list teams, list meetings, find person. And these two are the probably the most important. Get meeting summary and get meeting transcript. So let’s test the connection. So, let’s go in and find this call that I had with the team on March 6 and say I had this call, please use Fathom MCP and give me a summary. There we go. You can now see it says composing. And now you can see it’s calling Fathom right here. And there we go. Now we get a summary of the meeting and we can get all of the context and we can ask specific questions to pass about the meeting. How we built this, how we built that. This is extremely useful if you’re just starting out because then Claude have insights into everything and can help you build anything that you want that you say on the call that you can build. So if a client wants a website that can do some specific thing, you can have Claude read through this transcript, build the entire plan for you, and then build everything out from you directly based on the call. Very, very useful to know you should use this for every single meeting that you finish. All right, let me show you how to do the service delivery so you can start building cool stuff with Claude. If we do a small plot search and we basically just ask how many businesses in the US are still missing a website, what we will find is approximately 30% of US small businesses don’t have a website in 2026. Right now there’s approximately 36 million small businesses in the US, which means that we have around 10 million businesses without a website. And that’s also why I previously showed you these three levels of things that you can build right off the bat where website is level one. The reason for that is that what this entire business model is about is that you learn skills, you learn how to do something well, where businesses would rather pay you to do it than do it and find out how to do it themselves. And building websites is one of those things that still have a lot of perceived value. If people see a very nice website, they think, “Okay, that must have cost a lot.” Even to this day where AI is so good at building websites, there’s a lot of perceived value in websites. And that is also why it’s what we’re going to start off with building. Here’s an example of the website that you’re going to learn how to build. And I’ll even give you this template right here. So, you can actually build exact websites like this from one prompt. You can see we have this navbar right here that when we scroll that becomes like a sticky nav bar at the top. I really like this design, the choice of colors, the choice of fonts. It looks very, very professional. And what you’ll also be learning in this next part of the module is how also make these websites actually high converting because that’s one of the places where a lot of website designers, they mess up. It’s because they create websites that might look good, but they are not high converting. You can see there’s a very bright button right here that you would want to click on. And you also have a number right here that you can call if you want to call them straight away. You’ll also learn how to create these very, very clean animations directly with code and plot code. You can see this little plumbing animations right here from the pipe and this calendar animation right here. All of these you’ll learn how to create. You’ll learn how to pron plot to get a result that looks this good. You’ll also learn how to create these type of animations which are kind of like these cards where they stack on top of each other which looks very professional. Then you’ll learn how to split this into a dark mode that creates like a very nice contrast. And at the end you’re going to have this form right here and I’m going to show you how you can hook this website up to any CM that you want. This is a very nice website and usually web agencies can charge a lot. If I create a new chat and I ask Claude, “How much does web agencies charge for a nicel looking website?” The result will probably surprise you. Look at this. Depends on a lot what nicel looking means, but rough market ranges. For a solo freelancer on a template where it takes 1 to 2 weeks, you can charge 1 to 3k. If it’s a small boutique, then it’s 3 to 10k. If it’s a proper agency, it’s 10 to 30k. If it’s a mid-tier agency, it’s 30 to 80k. And if it’s a ward tier, like a big web agency where they really spend a lot of the time on the design, they can charge 80 to $250,000. And if it’s some of the very well-known brands, they can charge $250,000. And you can even see for my agency recommend to $10,000 for websites. So you can see how much you can actually charge. And again, to set your expectations, you’re not going to go out and close your first website to $10,000. But what you might go out is close a website for $500, which would still be a massive win with users starting out. And we can build all of this from inside of Claw Code. I’m going to show you how to build it. Of course, I have a skill that creates these websites in literally just like 5 minutes. I’ll show you how you hook this up to any CRM so the information that is captured when people click get an offer is actually being sent to somewhere. And I’ll show you how to make your website responsive so it looks good on both desktop and mobile. And I’ll also show you how you can host and deploy a website for free up to a certain amount of traffic. But usually if you’re doing it for smaller companies, the hosting would be free. You’re going to learn a lot in this next part. So let’s get into it. To build clean websites like this, you’re going to be using the build premium website skill. And again, you can find all of my claw skills inside of our completely free community. Just go inside of classroom, then learning hub, scroll down until you find my claw skills right here. I’ll also leave a link to this right below this video. This drive folder includes all of my claw skills because after you’ve installed this skill and you write slash build premium website, you can see this skill builds a premium animated marketing website with reactiv and tailwind CSS for any industry. Use when the user has to build a website blah blah blah. So we’re going to use it and write /bu premium website. I want to build a website for a landscaping company. Let’s just call it cityscape. Then we hit enter because what this skill will now do is that it’s going to force Claw to ask follow-up questions about exactly what side it is that we want. So the colors, etc. There we go. Now it’s going to ask us what the theme should be. I think a bold modern theme looks good. The brand color direction, urban slate, m forest, rust, olive, brass. I think this dark green one, moss green could look pretty cool. So let’s go with that. Let’s go with design. Heartscaping lawn trees lighting commercial for the services. I think that’s pretty good. And then I’m going to ask it to build it in this folder that we are currently in. I’m going to write submit answers. And now it’s going to build out the plan for this website. You can see it’s reading all these files from inside of the skill that it uses as a reference for how it should design this. Building this out can take a couple of minutes. And there we go. That took 8 minutes and 41 seconds. And now you can see site is live at localhost 5173. So, I’m going to copy this. Then I’m going to go into our browser, paste this in, hit enter, and there we go. Now we have it. Cityscape, home, services, approach, process, contact. Then we have request a quote right up here. Landscape architecture for the city. We design, build, and maintain. And now we have a clean website in literally just 8 minutes. That’s based on this template. And you can see we have this animation right here with these stock images, which looks extremely good. We’re missing one right here. So we need to fix that. Every discipline under one roof from a single tree broom to halfacre estate planned from scratch. This looks extremely good. Request a quote. When they click on this, they scroll to this section right here where they can fill in this information. They can even attach photos if they want to and then they can send the inquiry. Absolutely insane website. You can see this website right here is built from the same template that we used for the plumbing company. I’m really happy with how this site turned out. We can also right click and click inspect and then turn it to mobile right here. And then we can see, okay, it also looks good on mobile, which is also very important for websites these days. Great. You’ve learned how to create insane looking websites. And you can sell this for probably around 500 bucks. But let me show you how to also deploy it. You deploy it by first going to GitHub. If you don’t have GitHub already, then create a profile. Then click new. And I’m going to call this cityscape right here. You’re going to set it to public because you don’t want to give everyone your code to your website. And then you click create repository. There we go. And now you can take this URL right here and you can pass it to cloud code. And you can say push this code for the website only to this repo. Then you don’t want to have it in auto mode because it won’t be able to push to repo in auto mode. So you’re just going to set it like this. Hit enter. And now cloud is going to push all of this code into the cloud into GitHub. And you can see it’s first checking. Hit enter. When you do this the first time, it’ll pop up with an authentication where you have to authenticate. In order to do this, it’ll basically just open a window on your GitHub where you just have to click accept. But because I’ve done this plenty of times, it already has a connection. Now, it is creating a commit. And you can see pushed code is live at this GitHub URL right here. If we then hit enter on the GitHub repo, then we can see our code from this website. and hosting it is actually extremely easy when we have it on GitHub. I’m going to use the platform that’s called visil.com which is a hosting platform. Then if you don’t have it already, create a user. It’s free. Go to the top right corner, click add new, and then project. And then you can paste in this GitHub repo right here when you have your GitHub account connected to your VIL account. This makes it very very easy because VIL can now see automatically. Okay, this is a V project. It’s called Cityscape. And all we have to do is just click deploy once. And there we go. Now you can see it’s being deployed. This should take usually around something like 30 seconds before the website is deployed. This actually only took 9 seconds this time. You can see it says congratulations, you’ve now deployed a new project. We can then click continue to dashboard. And here we have our project. We actually already have a domain that’s given by VIL. If we copy this domain right here and we open it in a new tab, this is the domain that we’ve been given. Then we actually have a preview of what the site will look like when it’s live. And it’s actually live right now. We can go to this website from anywhere. And you can see it looks extremely good. Looks exactly like it did on development. If you wanted to add a domain, you would come down here onto domains on the left hand side. And you can see we have our standard domain right here. You would just click add existing. And here you can add any domain that you want from GoDaddy, from NameCheep, from Cloudflare, it doesn’t matter. It will take you through the review process of setting up the DNS. And when you have done that, then you have deployed your website to the web. If you wanted to sell these kind of websites, I would create five different templates and then create a portfolio website where you show these websites that you have built. That makes it much easier for a potential client to make a decision. Okay, should I work with this company or not? And if you’re using the method that I showed previous in this course, which is signing up for Upwork, then you can find plenty of work. For example, this one right here. Posted yesterday website for a construction company, logo and website design specialist, website design and launch. Looking to make a website. There are plenty of jobs in doing this. And this is one of the best ways to land your first couple of clients. The great thing about this is that when you have built up a portfolio and when you have built a bunch of skills like this, I’ll show you later in this course how you can create something and then create your own skill so you can replicate the process much much easier. your work becomes almost automated. Claude can do 95% of your product delivery. That’s the entire point of this business model. It’s to build stuff with Claude and then creating skills from it to replicate the process and making the service delivery so much easier. All right, that is websites complete. The next thing you’re going to learn is how do you actually automate things? If a client comes to you and says, well, I want my invoicing process automated or I want my onboarding process automated. How do you actually go about doing that? How do you create automations with cloud code? Usually, you would use a platform like make.com or like nm. But with cloud code, there’s a much better way of doing it. Now, we can use the platform that’s called trigger.dev. Trigger is an open- source platform for building and hosting automations. You can see that it’s open source right here. It has 15,000 GitHub stars. And we can see the entire codebase right here. Why is it smart to use trigger.dev over something like Naden or make? It’s a very simple reason. The reason is that trigger.dev is built with code and cloud code is really good at coding. We actually already have a skill that’s called trigger.dev. And this skill is going to be very important when we build out those workflows. What I’m going to do is that I’m going to clear the session inside of cloud code. And let’s build out a couple of automations with trigger.dev so you can see how it works. I’m going to write I want to automate my invoicing process. I want to be able to fill out a form with a bit of information and then it should build the invoice export as PDF and then send via Gmail to the client. Use two skills and you will have access to both of these skills as well. First one is the composio skill. I want to use composio for authentication. Skill two is trigger.dev. I want to build and host this with trigger.dev.

I’m going to give it an ultraink. Put it in plan mode. And let’s build this out. Remember what I showed you before with the process of connecting claw directly with your AI noteaker like Fathom via the MCP. Right now I explained what I wanted, right? But if you’re talking to a client, I want this and this and this automated. You can take that directly, that context directly from the call into cloud code and then you don’t even have to explain it like I did here. What fields should the invoicing form capture it should capture it all. So business information, contact information, line items,

tax and discounts, everything that we usually have on an invoice. It should create its own invoicing number and it should take the date automatically. Then we hit enter and there we go. It’s now going to plan out the process of automating this process of creating invoices. For how we trigger the automation, I think we should use a simple web form from Nex.js. So I’m going to choose simple web form and hit enter. For the environmental value, it’s very important that they’re not hardcoded but in av file. these let’s use react for the PDF and for the invoice number let’s do it date based and then we hit enter if you know how to build with cloud code it almost doesn’t matter what it is that you’re building if you know the process of first planning it out making sure that you actually give it the right context so it knows exactly what you want then the opportunity is literally endless you can build basically anything that you want if you know the actual formatting and the method of building stuff here it’s asking about the invoice counter I think it’s a bit overkill to set up the invoice counter in a database I’m just going to say let’s skip the invoice counter for now

and just do it date based the naming for the invoicing I think both attaching the PDF and storing it in Google Drive is going to be best who writes the subject fix template with placeholders let’s actually do it AI generated so the email is AI generated and then we hit enter now it’s going to plan out exactly how we can do this and there we have it we have an entire plan for how we should build this out. What’s worth noticing is that we are using Composio. If you don’t know what Composio is, it’s a tool for authentication. So, we can authenticate to a bunch of different things very, very easily. I made a full guide for Composio on my channel. So, if you want to, you can go and watch that later. And then it’s using trigger.dev for building and hosting this automation. The great thing about trigger.dev is that it’s open source. And when it’s open source, we can host this anywhere we want. The easiest thing is probably just to host it with trigger.dev and that’s how they make money as well. But if you wanted to or if your client wants this on their own servers, then you can literally just build these workflows and host them wherever you want. That’s the power of open source. So now we have a plan. Let’s take yes and use auto mode and let’s build out this automation. There we go. Now we can see that it is built. Let me open another terminal. The first thing we do is that we need to fill out this envir environmental file right here. I’m going to click new file. Add add ain file. Copy this example over. And then we need to fill out all of this information. So the first thing we need is a trigger secret key. If you don’t have it already, then make sure to create a user on trigger and log in. I’m going to create a new project. So in the top left pan, I’m going to create new project. I’m going to call this one invoicing and just click create. Now we can head under API keys and then copy this secret key right here. So we paste that in where it says secret key. And then we also need to go in and find the project ref right here. Copy this one as well. Now we need the Composio API key. So we go to Composio and Composio is basically what manages our authentication. If you don’t have a Composio account then sign up for one. It is free up to a certain amount of tool calls. Copy the composure API key and insert that as well. Now we need the enthropy API key. This is used for the AI generation of the email. So we’re going to platform.plot.com. Go into API keys. Click create key. Call this invoicing. Click add. Copy this key. Paste that in as well. We need the Google Drive folder ID. We get that by going to drive. We just create a new folder. I’m going to call this invoices. If we click in on that, we get this ID right here in in the header. So, I’m going to copy this one. Then, we write all of this information, which is our business info. So, let’s just try and see if it works. I’m going to write test street 1 2 3 building 4 the business phone number. Just write in some random phone number. Business logo URL. We can go to our shiny website. We can copy this image URL right here. That way we don’t need to host it, which is smart. Business bank details. Let me just write a random eban. Default currency USD. Default tax rate zero. Default payment terms net 14. Default due days. This looks good. Awesome. Then we close down this environmental file and we say great I have filled out the invent file. What now? How do we or now? Run the one time composure or script. It’ll print out two or URLs. Gmail drive. You click through both and it waits until each connection flips active. Let me just create a new API key. Maybe it was an old one that’s already rotated. I’m going to write try again. We still had a bug. This is a new bug. So claw can just fix that. There we go. Now you can see it says Gmail URL. So it’s going to give us this URL right here that I can copy and put into our browser. And this is how composio works. This means that we don’t have to set up the Google credentials ourselves. We can use composio that just gives us this link right here. And then click continue. And you can see now composio is connected to Gmail. I can now say done. Come with all URLs please. So we can connect. And for each service that we connect, it’s going to give us a new URL. So now we have the Google Drive URL right here. Paste that in once again. Log in with our Google account. There we go. And now you can see then open localhost 3001 server is still not running. So I’m going to ask please run the dev server for me. And there we go. Now the two servers are started both the back end which is trigger.dev and our front end. We can go to localhost 3001. And now we have send an invoice generates a PDF drafts with email withdrawal and ships it via Gmail from default currency USD tax 0% terms net 14 due in 14 days. So I can write the plan’s name is let’s do Albert testing and let’s do the company as let’s just call it the agent and the address we maybe don’t need to I can do like King Street 420 line items we can do consulting services one quantity price let’s do 1,500 we need another which is upfront fee let’s do one call that 500 currency money and date currency which be will be USD tax rate zero Discount type none. Discount value none. Issue date, let’s do that. The 18th. Due date, we can just leave that. Payment terms net 14. Any notes? No. And then we can click generate and send. Of course, we need an email right here. Let me just write in my own email just as a test. And then click generate and send. You can see it says cute run ID and then this run right here. Let’s see if it works. Let’s ask cloud now to check the logs. Did my test work? You can see it checked the logs and found a couple of issues. So let’s let plot fix those. Keep doing runs and you can see plot fixed one issue. Let’s try and check the logs again. Ran into another bug. We’ll fix that. Try and submit once again. Check now. And we just keep doing that until cloud has fixed all of the bugs. And you can see composer ID this depending on the tool. Submit again. Let’s do that. Generate and send now. There we go. Finish. Success. Let’s see how it looks. If we go to my Gmail and now hit a refresh, you can see we just received an invoice from me. Hi Alba. I hope things are going well at build my agent. Please find attached invoice for $2,000 dub by payment at terms of 14. So let us know if you have any questions, need anything from my end. Thanks so much for continue trust and working with shiny. Really means a lot. And here we have an invoice that is generated 100% by our workflow. It has all of our information invoice number like this and this and this. net 14 payment details. Boom. If a company now comes to you and say, “Well, we spent a lot of time sending out invoices.” You know how to fix it. And we’ve just successfully built a trigger. Workflow without even being inside of a workflow builder or anything. Claude did it all. But let’s say we want to deploy this, right? Right now, it’s running from our computer. It’s doing all of this from our computer. We can then write let’s deploy this to production on triggerdev. Thank you. The way that trigger.dev dev works in the cloud is that you have these projects right here and inside of here you have different environments like right now I’m in development but we want to be on production. So now you can see cloud is deploying this. It’s pushing it into the trigger.devcloud. What we have to do is just set the environmental values. We do that by going into environmental values right here and then clicking add new. Then making sure that we have it on production right here. A quick tip is that you can go inside of environmental value. Copy this. Then you can go back and paste all of this in. We’re just going to show all of the environmental values the right way. We can click save. And there we go. All of the environmental values has now been added. And there we go. Plot has now pushed our task. You can see our task inside of here now. Send invoice has now been added. We don’t have any activity yet because we just pushed it to production, but we can run it again if you want. So let’s test it. Before we test it, we need to take the API key from inside of production cuz that’s going to be a different secret key. And we need to go in and fill out this new information. So the trigger secret key, change that to prod. Same with the project reference right here. That’s actually the same. So we don’t need to do anything there. Let’s test it now by running npm rundev. I’m going to write let’s test it in localhost 3000 first form. I change the environmental values. Let’s try and test it now. So I’m going to write elot testing 2. Going to send it to my own email once again. I’m going to call it build my agent 2 and addresses we don’t want I’m just going to call it testing for let’s say 1 1500 USD blah blah blah all of this we don’t need generate and send now you can see it says sending cute run if you now go into trigger.dev and to our task you can now see one run was just cued right here we can also see run is now being executed. That’s one of the good things about trigger.dev dev as well is that all of this dashboard right here where it’s executing all of the runs, we have a very clean dashboard where we can see what’s going on. And now we can see right here the run succeeded in 13 seconds. If we go to our invoice and give it a refresh, we now have a new invoice with this new information and our invoice generator is now deployed in the cloud. So that is the back end, right? The backend is now deployed on trigger.dev. If we wanted to deploy the front end as well, we can also do that. We head to Vessel and then we write great let’s push the front end to Vessel. First GitHub though we’re going to go to GitHub and just like we did before we’re going to create a new repo. I’m going to call it invoice front end. We can choose an owner. We’re going to choose ourselves. We can click private and then we can create the repository just like this. Paste this into claw. Hit enter. And now it’s going to push our front end. Our front end is this page right here, right? The invoice generator. It’s going to push that. So, we have it online. The crazy thing about this is that you don’t have to remember, okay, how exactly did Elbert do this. You just have to remember, you can use Composio for authentication trigger.dev for the actual automation. When you tell that to claw, claw can figure out the rest. There we go. It is now pushed. We can now, just like we did before, go to Visil, click add new project, paste this in, click deploy, choose next, and just click deploy again. And that will deploy our project to the web. There we go. It is now deployed. We can click continue to dashboard. Click on the domain right here. And now you can see the invoice generator is live on the web. The last thing we need to do is just set the environmental values. You do that by going in environmental variables right here. Click add environmental variables. And just take the same one that we used before, paste that into the cell like this. Click save. You’re going to have to redeploy. And now the invoice generator is live on the web. What’s important to know is that right now we have hooked our entropic API key up to this invoice generator. It generates this text right here with AI when it sends the email. The issue with that is that if someone got access to this invoice generator, they can send maybe a thousand emails which will burn a lot of credits on Claude. So you want to add some authentication. You don’t want to just deploy it like this. And in the next part of the course, I’m going to show you exactly how you do that because you’re usually not only going to give them one of these like automations. Usually when you automate something within a business, you’re automating maybe 5, 10, 20 different processes. That adds up to a bunch of time. This is just one of them. In the next part of the course, I’m going to show you how you can add authentication as well, so only the actual company can access this. And I’m going to show you how you can build almost sort of a mini app with a bunch of individual small tools like this one where all of the automations are in one place that you can then give to the client, which makes you look very, very professional. And a collection of these is something that you can charge 10, 15, $20,000 for if you’re doing it for big companies. There we go. Deployment has been created. If we go to it now, you can see it has all of our information from before. And now it will work from the web. Throughout this course, we have built a lot of things using skills. For example, the website that we created was heavily carried by this build premium website skill. This is probably one of the most important parts about this entire course. And the reason for that is that one of the most important skills that you want to learn if you want to do this oneperson AI plot business is to create a process. Spend a lot of time building something out like this invoicing for example and then building a skill so you can cut a lot of time if you had to build it again. So learning how to build generic skills that builds out a process exactly like how you want it will save you a bunch of time in the long run. And the way you want to do this is that you want to use my skill that’s called create skill. This skill is also inside of our drives. this one right here called create skill which instructs Claude in how to create a good skill. I’m going to say nice, well done. This was a success. I want to create a skill for creating simple automations like this where I have some kind of form

and that is on next.js. That’s what we used and the back end is on trigger. Authentication if possible should be on composio. I want you to take what you learned from this session and add that information into a skill so we can replicate it in in the future. Very important. use the create skill skill that you have access to. The user should in the start answer a couple of questions about what it is that they want to build. The skill should be generic for any mini automation/tools that they want to build. So it shouldn’t be invoicing

specifically in the skill include that claude should research before building. Then you want to hit it with an ultraink. You want to hit it with plane mode. And sometimes something I also like to do is to use sub agents to go into our current app as the example and get all context you need. This is something that I like to write this use sub agents because claude is then going to speed up the process of finding context. Hit enter and now Claude is going to build out a skill for how to create automations like this. This will take a couple of minutes. There we go. Now it’s asking some question like what should the skill be named? I think mini automation fits pretty well. Should the skill bundle is start a template directory? Yes. Where should it live? Uh let’s do in the personal skills and then click submit answers. Because we use plan mode, it’s now asking us if we want to accept this plan. I think it sounds good. So let’s accept it. And Claude is now going to build out this skill for us. And there we go. After a couple of minutes, the skill is now built. Something that you always want to ask is, is this a standalone skill that doesn’t reference the invoice app? Answer this, please. The reason for that is that sometimes Claude likes to reference files, but if you want it to be its own skill, we don’t want it to reference this folder. We wanted to include all of the information in that one skill without needing any other context. But you can see it says yes, fully stand alone, no path references to any other skill. Awesome. Now we have created a standalone skill. When we write Ctrl C and then click plot right here and we now write mini automation, you can see we have a skill that’s called / mini automation. Build a mini automation in XJS form on the front end, a trigger.dev background task on the back end with composio for third party authentications, etc. blah blah blah blah. And now you can replicate the automation that we just created with that front-end form for anything that you want. and you want to get a habit out of doing this. So, anytime you build something that you think, well, I might build this for another client in the future, please make a skill for it. It’s going to save you so much time. And especially if you’re building out a websites and you want like a couple of different templates to choose from, building out a custom website and then making it into a skill is going to be so valuable for you. I personally believe that in the future with these Claude businesses, the value, the IP is going to be the SOPs and processes that Claude knows just like in a regular business. Now, what’s really valuable is all of the documents that businesses have that explains how they do things. The value in a Claude code company like this will be the processes, the skills that Claude has access to, the information that you have given it. So, please make sure to make those skills. They’re going to become valuable and save you a bunch of time in the future. Make a habit out of it. It’s one of the most important parts of running a plot code business like this. All right. Now, we have gone over how you build websites for clients, how you also build automations and how you can include AI into that to build agents. The third thing we’re going to cover is kind of like the next level to that, which is how do you then build out full AI systems? How do you take a business and then automate as many processes that you can and then deliver that to the client? That’s what we’re going to cover now. A full AI system is basically just a bunch of automations combined. So just like the invoicing automation that we just created, when you provide a full system, you usually automate maybe like 5 to 10 to 25 depending on depending on how big the project is. So how do you go about delivering this? This varies from agency to agency how you actually do this. Some agencies just deliver a bunch of init flows where if the client wants they can look at the executions of the init flows, but the reality is that the client is never going to do that. So the client just kind of trusts the agency that it’s set up correctly. But that is how most agencies use to do it. When we’re building with clawed code, we want to give Claude access to everything as well. So we’ll do this a little differently. We’re going to be building a dashboard for the client so they can see exactly what’s going on, exactly how many executions, maybe even exactly how much time these automations have saved them so far. That also allows us to provide them with tools like we just created with the invoicing tool where if they have some process that requires like something to trigger it like a form for example like we just showed you can put those tools inside of this system as well. This also allows us to have a front end where they can see all of this and then also have a separate backend where everything is running and depending on the size of the project and the client that you’re working with. Sometimes the client wants to run everything on their own servers and this system is great for that because the front end, the actual visuals and the back end is all going to be running as code. And specifically the back end, we’re using trigger.dev exactly like we did in the previous automation. And because trigger.dev is open source, it means that we can host it anywhere. We can host it on a virtual machine like a machine that’s just in the cloud. We can host it on the servers that they are already running. If they’re using something like Azure, which is the most normal one, that is like a cloud provider it’s called, or if they’re using AWS, Amazon Web Services, or if they’re using Google Cloud. Having a system like this means that you can basically deploy it anywhere you want with a very nice user experience. Let me be clear from the start, sometimes it’s overkill building out a full AI system. Let’s say that you’re delivering a project for a client and all they need is just like one automation done. What you could do is lally just provide the tool like we just did connected to something like a type form where you don’t have your own front end. This is for bigger projects where you are auditing the entire business and then say okay these things are what we can improve and I’m going to build this out like this and this and this and you’re going to basically get your custom dashboard, custom app where you have access to all of this. All right, let’s get into how we actually build this. So first thing we’re going to build out is the structure of what is really going to be a full app. We’re going to have the front end and this is going to be very similar to what we already built. It’s going to be next.js as the framework. It’s going to be tailwind CSS for the design. And then I like to use something that’s called Next O for the authentication that works well with Nex.js. And then we’re going to use resend together with next off. And resent was basically the service that we can use to send emails like magic links. And this is the most secure and easiest way to set up a login. That way you don’t have to manage passwords. You don’t have to manage usernames. All you’re doing is just sending a magic link. Every time someone wants to log in, they click on that link and they’re in. So it’s basically like forcing a two-factor authentication every single time, which is the most secure thing that you can do. Then we have the backend. This is where all of our automations are running and we’re going to be building that in trigger.dev. We’re going to start by hosting it on trigger.dev. just know later on if you wanted to host the back end yourself, if you wanted it on something like a virtual machine or hosting the back end on the client’s actual servers, then trigger.dev is the perfect thing to build with because it allows you to just take the code and just host it somewhere else as well. And then because we’re building out kind of like a full app, we also need a database. But the database I like to launch fast and the fastest way that I found to launch is by using [  ] DB, which is a database provider. And the database is where we are basically going to be storing all of this data. So all of the invoices for example if we have an invoice generator, all of the clients information like the email and user information. And what’s worth knowing as well is that we also going to be hosting MongoDB in the cloud just to launch fast. If you at one point wanting to build this entirely on the client’s platform, then you can also do that because MongoDB also has a community edition that allows you to self-host it just like you can with trigger.dev. And while it’s not entirely open source, their community edition is fine for what you’re building as well. So that is kind of the stack that we are going to be using. And don’t worry if you don’t know what all these things are yet. After watching this part of the course, you’ll know exactly how to set up full AI systems like this. All right, let’s get into it. The first thing we’re going to do is that I’m going to build a new folder right here. I’m going to call this one, let’s just call it AI system. And then I’m going to go out of claw and write cd AI system, which is basically just going to get me into that folder. and then write plot. The reason I do this is because I don’t want to give it the context of all the other things that we have built as well. We want it to be completely focused on this right here. What I’ll then do is that I’m going to write I want to build out a full app for a client that includes a bunch of small automations. And then I’m going to write the only domain that should be allowed into this app should be at shiny AI domains. The reason we do this is because if we deliver this for a client, we want to add some additional security measures that only emails with their own domain should be allowed into it. This is basically just an additional layer of security. Then I’m going to write the stack next.js. We want to use next off for authentication with no Google login because we don’t need that. We want to be using the magic links, but again, this is something that you could add later if you wanted to. Then we want to use Tailwind CSS recent for magic links. For the back end, I’m not going to write yet. I’m going to write backend. Let’s not create this yet, but I want to use trigger.dev at one point. So just create the structure for its and then for the database we want to use [  ] I’m going to give it some additional information. This app will be a dashboard for the automations they have running. So keep as much as possible in the front end, only automations at the back end. Only automations and processes in the back end. For database, we’re going to use MongoDB. And then I don’t want to build anything out yet. I don’t want you to build out any automations yet. I want you to create the structure of the app. I want a dashboard when they land inside a simple sidebar where they will have all of the automations and a settings tab where they can log out etc. Please build this out. Then we’re going to use ultraink. I’m going to put it inside of plan mode to force it to create the full plan before it starts building. And then we are going to hit enter. It’s going to take a little longer now the first time that we are building it out. But just like we have done previously with the automations, what we’ll later do is that we’re going to create a skill out of this. And I’m going to give that skill to you as well. And then we can basically build out full AI systems like this in a single prompt. Which next off version? Use the latest one. For components, we can use shared CN components. How should the shiny AI domain restriction work? only allow add shiny.ai domains MongoDB persistence for all. Let’s use the JWT session tokens and submit answers. There we go. Now it’s going to plan out this entire thing. And there we go. Claude has now written up a plan. I’m also going to tell it to please use the super powers skill and I approve of plan. Again, if you don’t know what the superpower skill is, superpowers is basically a plugin that includes a couple of skills for like writing plans, writing skills, executing plans. It’s basically forcing Claw to think like a developer, which is very useful when it’s building out a full app. So now it’s going to build out and scaffold this entire project that took 13 minutes to build out. Now we want to click into the folder where we have created all of this. We want to go into the end local example, copy this, paste this, and then rename it and just call it.local like this. Cool. Now, we need to fill out all of this information. The first is the or secret. To create this, we can use this command right here that it was so nice to show, which is called open SSL rand and then B 6432. Hit enter. That’s going to give us a random string of letters. We can paste that in instead of the O secret right here. Just like this one. Then we have the next O URL. For now, we’re just going to keep that at localhost 3000. And then we go under API keys, create an API key. I’m going to call this one AI system. It should have full access. Click add. Copy it. Go back. Paste it in right here. And then I’m just going to write shiny portal. That is going to be the sender. And for the email, I’m going to use this email that I’ve already set up, which is just going to be this one. So, it’s going to be no reply at mail.clipip.dev. This is the domain that I’m just going to use because I’ve already set it up once. And we’re using the MongoDB URI. We get that by going to MongoDB. If you don’t have one already, then create a user. Create a database and a cluster if you don’t have one already. I’m now going to click connect. Click drivers. And this gives me this string right here. What we then also want to do is that we want to click done right here. Then go into database and network access. Click edit. And then we can edit the password. also generate a secure password. Copy this and paste it in right here where it says DB password just like this. And the trigger.dev key is not set up yet. We’re going to set that up later when we set up the automations and the back end. Now we can click update user right here. And now it should work. The last thing we need to do is that we need to insert the name of the database. It’s not called shiny right here. It is called newsify right here. It’s from a pre previous project. So I’m going to fill this out as well. Cool. Let’s try and run it. So, we need to go inside of AI system by using cd and AI system. And then we’re going to write npm rundev to start the dev server. Hit enter. This is going to open it up. And you can see it starts on localhost 3000. Let’s see what it build out. We definitely need to work on the design of this. Let me sign in with [email protected]. Send magic link. There you go. Check your inbox. Can then go inside of our inbox. Elbert shiny.ai. Click sign in. And there we go. Now we have our dashboard right here that shows total automations, how many active, how many drafts, recent activity, nothing yet. Welcome back. Your automations dashboard, blah blah blah. And then it says shiny up here. Then we have an example automation right here where we’re going to have all of the automations here on the left side. And then inside of settings, you can see you signed into shiny email and then you can sign out. Great. This is just the structure of the dashboard. We definitely need to make this look a lot better before we do anything else. The way we do that is that we use the front end design skill. I’m going to write that looks good. I want to give our app a makeover. Use the front end design skill. I want it in a linear light mode type of vibe. We should have both light mode and dark mode. Let’s build this out. Thanks. Ultra think hit it with a plane mode so it plans out how we should do it and then hit enter. I like to build the structure first where it doesn’t care at all about the design first so it builds it out the right way and then later we give it a makeover so it doesn’t look this bad. That will probably take a couple of minutes to fix. Great. Now we have something that looks more like this which looks good. But I can see we already have an error right here. So I can click on this little icon to copy the error. Go back into claw, paste it in, and say, “Please fix this. Thank you.” And now it’s going to fix this error right here. It’s running a lint. Go. Let’s see if it There we go. The error disappeared. If you write at [email protected], you can see it says check your inbox. We can then go to our inbox. Click sign in. Now it says something went wrong, which is weird. That’s because I clicked on the one that was 20 minutes ago. You got to click on the newest one right here. 0 minutes ago. And there we go. Welcome back. Now it looks like this. And the light mode. This looks a lot better. There’s some weird thing going on with the gradient. What I’ll do is that I’m going to take a screenshot of this and say this title and other text has a weird gradient. That is too much. Please fix it. Thanks. I like this theme a lot better. And now we have both a light mode and a dark mode. It fixed this elbow text right here, but it still does that weird thing. I’m going to take a screenshot of it again and say, “See how in the sides it’s less bright than in the middle. We don’t want that. Please fix.” Also, this light mode button up here. We probably don’t want that. So, what I’m going to do is that I’m going to say, “Nice, the light mode button top right corner. Please remove that so we only have it inside of settings. Looks a bit brighter, but it still has that weird shadow on the left hand side right here. Let’s try and write on the sides. There’s a weird shadow. Can we please remove that? Actually, I think it’s because these dots in the background. They probably need this shadow. So, I’m going to write they should only have the effect on the background dots, not on the headlines and text and components, etc. You can see it’s doing it on every single page. Kind of like dark to the sides, but then in the middle it’s like bright. We want it to be white everywhere. There we go. Now, it fixed it. So, the text is completely white. I think this looks a lot better. What we’re going to do now is we’re going to install our first app. So I’m going to write let’s install our first app which will be this invoicing app that also uses trigger.dev as back end. I want you to one install this front end form to our AI system folder and two create the trigger.dev back end. We’re going to call this the AI system back end right here inside this folder that will include the back end for this automation.

also make it so we later can add more tasks. That’s what it’s called inside of trigger.dev. Uh so more autoations later. For now, let’s add this one. I’m going to set it to plan mode and write use sub aents to get all context you need and implement this automation. So now we’re going to take the invoice app that we just built previously and we’re going to install it into our AI system. And there we go. Now you can see inside of our app, the send invoice automation has now been set up. And now they have the same form that’s now in this new branding where they can generate and send an invoice much faster. And then when you’re building out this AI system, it will consist of let’s say five to 10 different automations that are automating something. And all the employees would need access to is just this dashboard right here that would allow them to use the automations. The next thing I want to do is to handle the authentication. If it’s not off yet, I want to add a off button to this right here. For this automation, we need to handle the off. So, please add if we don’t have authentication set up when we trigger it that a modal pops up where they can authenticate. When we created this standalone automation, we handle the authentication using the link inside of the terminal via cloud code. But of course, our clients will not be able to do that. So, we need to add a nice button where they can just authenticate with whatever they’re using. Cool. I hope you see the idea of how you can build fully custom AI systems, how you can make it domain specific so only people from this company can use it, and how you now have this structure to build anything that you want inside of here. Your automations doesn’t have to be a form where then something happens. You can also set up automations that doesn’t require anyone to do anything where it just runs on autopilot. Let me show you how to do that too. But before we do that, let’s make sure that this actually works. So the first thing I’m going to do is that I’m going to go to trigger.dev. I’m going to create a new project that I’m going to call AI system. Then I want to push this straight to production. So I am going to get the API key right here, the secret key, paste that into environmental value right here. And we also need to do that right here. Then we need to go down and we need to get inside of general the project ref. So I’m going to insert that here too. Close this down like that. Just like before, we need to set the environmental variable. So I’m going to go inside of here, click add new, and then go inside of our environmental file. Copy all of this and then paste this into the production like this and click save. And now I’m going to wait for it to set up the connection so everything works. And then we’re going to test it. Cool. Let’s try and push this to production. Going to give it the project ref right here. I’m going to say we need to set up the back end. I have installed all the N1 trigger.dev. This is the project ref. Can you please set this up? So it works with everything authentication, etc. Thank you. So now we’re basically just going to push this backend code right here, which is the AI system backend. We’re going to take these tasks that are the trigger.dev tasks. We’re going to push it to this new trigger. That we just set up. So we have everything in one place. You can see there’s no tasks inside of here yet. And there we go. I now deployed it to production. Now you can see the task is inside of our system on trigger.dev in the production. If I then just make a test, we can call it Elbert test. company name BMA just write consulting set it to $1,500 and click generate and send you can see it says there we go cute run ID run like this and now if you go inside of runs you can see it’s now executing and you can see invoice from shiny here we have it cool so now our send invoice is working what I’ll do now just like I explained before is that now we have the start right this is a good starting point for every new client that you get on as well this overview dashboard where you have a company name and one automation is the perfect starting point for every new client that you sign on as well. So, we want to give Claude this information. What I’ll do is I’m simply just going to create a new session. Then I’m going to use the create skill and I’m going to drag these in and I’m going to write I want to create a skill that creates this kind of front end and backend system to when I close a new client. It should be a generic skill. So first it should ask what the client name is, what the domain is for their emails, etc. It should gather all of the context first. Then I wanted to build out this template with the front end and back end first. After we have built out this structure, we should not have any automations yet, we will then later build automations, but the entire structure should be there. So it’s very easy to do. This skill should work stand alone. Should not reference these folders. It should have everything included with references etc

inside its own scale. I’m going to give it an ultra think and hit enter. Actually before I do that I’m going to write use sub aents to gather all the context for how this works. Thank you. I’m also going to write include everything like design, how it works, uh domain, the stack, etc. Use sub angel to gather all the context for how this works. Thank you. There we go. Bit of a long prompt, but we’re going to use the create skill now. And again, this skill is inside the claude skills folder, which is inside the free community. So to get this skill, just go inside of classroom, go inside of learning hub and inside of claw skills right here, you have all of my claw skills in this drive folder and it’s completely free to get. There we go. Now it’s going to start building out this and it’s calling it new client system skill. This will probably take some time because it needs to gather all of the context first and then build the skill. But you really want to learn how to do this because every time you build something new for a client that potentially could be replicated in the future, you want to build a skill around it. If you don’t do this, you’re going to start from scratch every time you land a new client. When you do this, you’re going to make your life much easier. And there we go. Now you can see the skill is ready. If I go into a new claude session, let’s restart cla right quick. And we then write new client system like this. It says scaffold new clients full stack next.js 16 front end dashboard trigger.dev bag and worker from the shiny automations template. Use when the user says new client onboarded client scaffold new client system, etc. And of course, I’m also going to drop this skill together with the rest of the skills inside of the Google Drive. Now you know how to create full agent dashboards and you can of course customize this to however you want. If you want graphs right here showing you the task, you can ask Claude to build it. If you want to add a specific automation, you can ask Claude to build it. If you want to add something inside of settings where they can add the users themselves, you can tell Claude to build it. But what I want to do now is build some more automations inside of this dashboard to show you how that would look as well. Let’s get into it. All right, let me show you how to build out some more automations inside of our own automation system. What I’ll write is I’m going to take this folder and this folder right here which is basically all the files that this system includes and then I’m going to write read how this works use sub agents and then I’m going to say specifically look at how the automations are created etum then return to me. We just want to give the context for how this AI system works before we then try and build out a new automation inside of this session. So it’ll probably take a couple of minutes. All right. So let’s automate some more business processes. Let’s say the onboarding for example. A lot of businesses right now they do manual onboarding. So they send out emails manually. They send out the contract manually. They do all of this stuff manually. We can create an automation for this as well and display it inside of here. So we can now write I want to create a new automation that is called start onboarding. This should let’s write up what it should do. Send over contract via Gmail using Composure of course. Send another email with a Calendarly link where they can book an onboarding call. Send an email with an onboarding survey for example. Send another email with Kennet link. So the connection we need is probably just Gmail. The user will provide onboarding link and K link because it already now knows the structure of our system. This should be extremely easy to set up. I’m also going to write please add this task to trigger.dev of course like the other ones straight to prod. And there we go. Now it’s going to build out this automation just like the other one. There we go. That was extremely fast. We can actually see that building out this automation actually took only 3 minutes and it was also a pretty simple automation, but we’re going to start here before we build something a little more advanced. But let’s say that we just closed me. This is my email. This is my company. Then we can say send a contract URL. This could be something like a docu sign. But we might want to change this to an actual PDF. So maybe let’s do that. Let’s copy this and say, can we instead here upload a PDF on the new tool. Thank you. Then it’s going to change the automation to do that. Then we insert the onboarding survey and then the calendarly URL. So the calendar URL could be something like this. The onboarding survey URL be something like this. And then we need the contract URL, but that’s going to change. Now we can instead upload a PDF, but I actually want to generate the PDF instead. Can we actually instead generate the PDF? For now, we just generate a dummy contract. But when we generate, we insert the actual name and business information, etc. Please update. And again, if the client comes back and says, well, I want this change, so can we do this instead? Then you have everything in code. B can literally change anything inside of here. So, it’s fully customizable. There we go. So now we don’t have a field to upload the contract, but now it’s basically just going to send out a dummy one. So let’s test if it works. Open a shiny AI BMA. Insert our onboarding survey. Insert our calendar form. Send this in. Click start on boarding. There we go. You can see it cued this run right here. If we go into trigger.dev and look at runs, we can now see that it’s now executing this run. There we go. It took 12 seconds. Let’s see if I received something in my email. Now I get the book your onboarding call, onboarding survey, and your contract. Let’s get started. This is all automatically sent. You can see we even generated this draft or placeholder contract that we can always change later to exactly what we want. But now it sends out this contract as well that includes my information. Absolutely insane. We just automated the onboarding process as well. And if the client says, “Well, we want a new lead updated inside of our CRM or we want something else done.” literally just tell claw to build it and it’s going to build it out for you. But let’s build out some more advanced automations. Let’s get into it. All right, now you know the basics of creating these automations. But let’s say that we want to build something a little more impressive. And this is usually also the things that you’ll be sending to clients. For example, let’s imagine that a client that you’re working with, they have five support reps that are answering emails all day long. and you want to automate, let’s say our goal is 60% of that. So all of the questions that are the ones that are asked all the time that an AI can literally just answer those questions almost instantly. We can also build a full AI agent system like this inside of our dashboards. The great thing about this is that it is custom code. It means that there are literally no boundaries. We can build whatever we want. So I’m going to go back into claw code. I’m again going to tell it to read the structure of this full AI system. Then return back to me, use sub agents, understand how it works, hit enter. It will gather all of the context that it needs and then it’s going to come right back. Great. It returns back after 2 minutes and has all the context. Then we can write I want to build another more advanced automation. This should be a full AI support email ticket response system that can respond to around 60% of all emails. In this automation, I want the front end to show the AI responses. I want it to show if any has been escalated. I want to be able to upload a full knowledge base. So, it should use rag to search through it and answer questions based on this knowledge base. It’s getting a little more advanced here, but luckily we do have claw that can build out all of this for us. We will use composio to answer the emails uh using Gmail. I want to trigger the trigger.dev every 10 minutes to check for new emails and then respond to every single one that has come in. If the ticket requires human attention, we should be able to escalate it and see the escalated emails in the front end. and be able to respond to them from within the front end as well. I also want to be able to see the AI responses inside the front end, too. Let’s see if there’s anything else that we need. Let’s plan this out. Please use the superpower skill and ask me if you have any questions. Let’s fire away that prompt. And you can see it’s going to load the superpower skill and ask us any questions if it needs more context. We can all of a sudden start building some really cool advanced stuff based on the current structure that we have. How should the AI reply behave for the 60% it’s confident about it should also send immediately. Of course, if you’re building this out for a client, you would of course have talked with the client and asked how it should do it. But you’re going to want to send immediately. What format will the knowledge base come in? Let’s do let’s do PDF files. Let’s actually just do text files. That is a lot easier. Well, we can actually turn on all of them to be the answers. Where should we store the vector embeddings for rag? We could use something like pine cone or postgress, but we actually already have vector search inside of MongoDB. So, let’s use that. Let’s do if any human asks for like low confidence from AI and tone and refund. And let’s also do specific clarifiers like this. Submit. You can see it’s asking all of the right questions that it needs in order to build out this plan. Which email account does this monitor reply from? One shared inbox. Yes. And there we go. Now you can see it’s exploring the project to get context. And then it’s going to ask a couple of clarifying questions. Come with a couple of approaches and then design the plan for building this out. How should we identify a ticket? What groups emails together? A Gmail thread is one ticket. That sounds good. When you escalate and reply from the front end, should the AI help draft the human reply? Yep. If we can get AI to pre-draft the reply but just not send. I think that could be pretty cool. Where do we store tickets, messages, KB chunks and embeddings? Reuse existing MongoDB. Let’s do that. Anything important I should add to the spec. Pick anything that apply per ticket tags, labels, analytics view. Could do an analytics view and a per ticket tag. Submit this. Submit the answers. Got it. And off to design. Let me present it in sections. Confirm after each section one. This is this. That sounds good. Then it’s going to use trigger.dev to pull every single 10 minutes. This also looks good. There we go. That was all of the sections. Now it’s saying writing out the spec docs. Now that’s probably going to take a couple of minutes. Now we have the full design spec, which looks good. So I’m going to tell it to continue. Now it’s going to use this design spec to build out the entire plan so it can build this out. It’s going to ask us if we want to use sub aent driven or inline execution. We definitely want to use sub agent to speed it up. This means that different agents are going to be working on different tasks. You can see I dispatch a fresh sub agent per task review between task fasteration usually the fastest and also what’s recommended in this superpower skill. So now it’s going to use the sub agent driven development loaded this superpower skill and now it’s going to build out this entire thing for us based on our instructions. This will probably take a bit to build up because it is a pretty extensive AI system that we are building out. There we go. The next thing we need to do is that we need to go in and get our MongoDB secret. This one right here. Then we need to go to our production trigger.dev. Go inside of environmental variables. Click add new. Click on production. Make sure that we have added both the MongoDB URI and the DB name. Click save. There we go. Now I’m going to write we need this to update on prodrigger.dev.

So it will work. Can we please do that? Hit enter. And now it’s going to push the changes it has made to our back end. It’s going to push those trigger.dev so everything will run smoothly. All right, there we go. If I now spin up the server, you can see on localhost 3000, we now have another automation that’s called AI support. And here we have a couple of tabs for escalated support request, AI replied, human applied, and all. We also have this analytics tab right here that shows us how many tickets we have. And we have this knowledge base tab where we can either paste text into the knowledge base. We can add PDFs or URLs. Let’s see if this actually works. I can take our shiny.ai website for example. Paste this in. Click add to knowledge base. It is processing now. Let’s see if it actually works. Says error though. So this didn’t work for some reason. Let’s try and see if we can fix this. I tried to add knowledge base. I ran into this issue. Can we check what happened? In the meantime, let’s try and check the other things like paste text. If I just like copy this entire page and paste it in right here, add to knowledge space and call it basic info. Let’s see if it wants to add this. But then give it a refresh again. Still says failed. So there’s obviously something wrong with the knowledge base, but cl of course be able to fix this. Ah, now I see why it fails. We need an openi API key. So I’m going to set this value as well inside our environmental value right here at the end openai API key. And then I’m going to go to platformi.com login with my Google account. Let’s just call it AI system. Create the secret key. Copy it. Paste it in right here. And then we also need to go into trigger.dev inside of the environmental v variables here. Click add new and add it here too. Let’s do a refresh of this right here. Go to support. We’re actually getting emails now that is escalated to us. No emails has been replied to yet. Let’s try and add the knowledge base once again. Copy our entire website right here. Paste it in. Call it basic info. Add to knowledge base. Now it is processing. Refresh. Still processing now. So it didn’t fail instantly. That’s good. Let’s delete these others while we are at it. And there we go. Now it says ready. So if we go back to inbox, we can see it has no subject line right now. So I’m going to copy this and I’m going to paste this in and say that seems to work. issue now is that the subject lines don’t show even though we do have subject lines. Can you fix please hit enter and then hopefully we can fix this subie line not showing but this is an actual email from my inbox which is kind of cool. Same with this one. This is also an actual email. So it’s actually fetching the emails which is kind of cool and it is escalating them. Great. I think I also want to add a refresh button somewhere on this page. So, I’m going to give it this page right here and say, can we add a refresh button here as well that triggers a polling, please? That way, we can also do it manually just to get all the newest ones. And there we have it. Now, we have a little poll inbox button right here. If we pull it, it says refreshing just like this. And then it shows us if we have any new emails in our inbox. So, let’s actually try it. Let me go on this email account right here. Write to my personal email right here. Basically, write what does shiny AI do? We can basically just ask, I saw your website. What is it that Shiny AI does? We can click send. And now I receive my email inside of my inbox. Let’s see if we also receive it or if we hit little polling right here. For some reason, it doesn’t show. Maybe it’s because it’s under AI replied. Hm. Check the latest poll. I ran it after sending an email to myself to check it, but the email doesn’t show up. Why is that? Check the latest logs. Let’s see if I can figure out what happened. There we go. The issue was apparently that I clicked in on it, so it became red. Of course, we only want to take the unread emails. So yeah, the system is actually working like it’s intended. Cool. But it’s still being escalated. So we need to fix some prompting. I think I can write, see this email. It’s a simple question and we have it in our knowledge base.

Why did it get escalated? Hit enter and see what it says. All right. This was another one of these issue where it wasn’t really an issue. I just had to refresh. But now I’ve added auto refreshes every 30 seconds inside of this tab. And if we go to AI replied and we click in on this, we can actually see, hey, saw your website blah blah blah. And then the AI actually replied with this. Hi, thanks for reaching out. China AI helps businesses boost sales. It’s using the information that it got from the website. And if we click why this reply, we can even see the confidence score of how smart it thinks it was. We can see the citations that it used from our knowledge base. This is extremely cool and it works. So now we basically have an AI inside of our support email that can reply to questions. When this confidence score isn’t as high, it’s going to put it in the escalator tab so we can reply to it. I hope you start to see the power of these insane automations and agent systems that you can all build inside of your own kind of dashboard right here and provide to a client. And the great thing is that as you build these systems out, you could then sell this support inbox to another client and it wouldn’t really take you that much work to set it up with a new knowledge base on a new client. And the crazy thing is the amount that you could sell this for. If you sold this to a big company, you could probably charge anywhere from $2 to $5,000 a month depending on how big the team is. Cuz think about it, right now their support rep is probably being paid maybe like $2,000 a month. If it’s in the US, then probably three to $4,000 a month. So if a team has, let’s say, 10 support reps and all of a sudden they only need five, you’re saving them five times, let’s say, $3,000, that’s $15,000 a month you’re saving them. And if you’re charging $5,000, you’re still saving them $10,000 a month. And this is just one of the automation systems that you can build out. You can automate all processes inside of a business. Smack all of these automations here on the left hand side, and you have a killer product. And you can manage all of this with cloud code. What I’ll do is that I will take the code for this entire dashboard right here with the trigger. Back end. So these folders right here, AI system and AI system back end. I’m also going to take the cityscape website. So you have that as well as reference. And I’m going to put the GitHub repos right below this video. The last thing I want to do is that I want to host the front end. Right now we are still in localhost 3000. The back end is hosted on the cloud.trigger.dev, but the front end isn’t hosted yet. And it’s honestly quite easy to do. It’s the exact same thing we did with the other website. We just open visil.com and click add new project. Then we want to push this to GitHub. So I’m going to say I want to push only the front end. So the only this AI system folder right here. That’s the front end, not the back end, only the front end to GitHub. Then we’re going to open a new tab. We’re going to go to github.com. We’re going to create a new repo. I’m going to call it AI system front end. We going to make this private. Create repository. Then we can copy the link right here. We can paste it in and say push to this repo. You won’t be able to do this in auto mode. So I’m going to set it to just like the default mode. Hit enter. And then we will have to accept the commands. It’s asking us what to push. I’m going to ask it. I want to push this to separate repo so I can host the front end. So only the AI system standard push it’s it’s empty brand new and then we push it. Something that I always like to do is that I write of course don’t push any secrets etc. Just so we don’t push any of these secrets to GitHub for security. Always good to remind Claude not to do that. Sometimes it gets ahead of itself. It is running all of these commands. We’ll just let it do that. There we go. Created a little mini plan for itself. Now it’s going to push this frontend repo to GitHub. There we go. Now it is pushed. If we go onto GitHub, then hit enter. Now you can see we have all of the files inside of here. Let’s host that by going to VIL, copying this GitHub repo, pasting it in, clicking deploy. It’s going to see that it’s a NextJS application right here, click deploy again, and let’s see if we have any build errors. See, we already have a build error. So what I’ll do is that I’m going to copy this. Better thing to do actually is to go inside of here where we are running it and then instead of npm rundev we can write npm run build which is going to show us all of the build errors inside of this folder. You see it compile successfully running typescript. This actually built successfully. So I’m not sure what this error is. So let’s copy this paste it in and say when I try and host on this I get this why the front end installed MongoDB7 but PN install both villa doesn’t pass legacies by default to clean fixes are going down MongoDB6 let’s do what it recommends there we go it pushed the changes to GitHub so let’s try and deploy it again we can just click deploy once more and now we should not get this error again you can see It passed the first 10 seconds, which was where we had an error before. And there we go. Congratulations. You now deployed a project. Let’s continue to dashboard. We get this domain right here that we can go to. And you can see this will now be the login screen right here, which will only allow us to sign in with a shiny domain. So, I’m going to write my email, click send magic link, and we see something went wrong, which is because we don’t have our environmental var variables yet. So, go inside of here, and then we need to fill out all of the environmental variables. Click add. Pasting in all of these. Click save and do a little redeploy like this. This will take another minute or so and then it will deploy it. Let’s click right here. View deployment. Look at logs right here. Build logs to see if it deploys like it should. There we go. Build completed. If we then give it a refresh and write in our email. Now it should work. I can see right here that it’s going to localhost and that’s because I forgot to change one of the environmental variables. So, let’s go to next or UL right here. You can see we’re going to be using this URL instead. Pasted this in, click save, and click redeploy. Right now, we’re using this domain right here, which is the domain that Vasil gives us. But you probably want to add your own domain when you set up an AI system like this. The way you do that is that you just go inside of VIL again, go down to domains, and then just click add existing, and you can write in whatever domain you want and connect it using DNS. If you have any issues with that, then you can just ask Claude. After you’ve changed the domain, then it’s very important that you go inside of the environmental variables, just like I forgot to do now and change this next off URL to be the URL where your app is running. If you don’t do that, you’re going to run into the same issue that we just did before. Now it’s deployed once again. Let’s hit enter. Write in elot shiny.ai. Now it is sending the magic of URL. And now we are in and we have our AI support. We have our AI send invoice. We have our start onboarding. We have all of the processes that we built just before. There we go. Now you’ve deployed an AI system that could be for a client. Congratulations for reaching this far into the course. Great. Now you know how to build in these three levels. Websites as the easiest thing to build, then individual automations to automate stuff with trigger.dev and then how you can build a full app and dashboard around it that has multiple different automations in order to deliver a full project. What I’ll show you next is my structure and my way of basically being able to build anything. Cuz now you know a couple of things you can build and deliver to a client. But if a client has some specific request and you want to build something that you don’t have a tutorial on that doesn’t exist on YouTube, how do you go about that? How do you build anything? I usually break it down into four different steps. The first thing before I even start building anything, we need to create a really, really good build plan. This includes things like the tick stack that we are using, how it will work, what database, what frameworks, what exactly is it that we are building. It also includes things like the design, how should it look, what should the feel of the thing we are building, what should that be like? And that is actually the most important step of building anything. And the reason for that is that there are so many things that goes into this step of figuring out okay what are we building and how do we build it specifically with the tech stack because if you’re building something like an app or an automation the text stack is what determines what the cost of running it will be and how fast it’s going to be as well and how well does it scale and if we at one point later on wants to add more things is that possible can we add more features can actually do the things that we want can we give it the functionality that we need and wants in our app or automation or whatever it is that we are building this build plan what most beginners do when they’re building something is that they just write I want to build this and this and this hit enter and then take it from there but what happens most of the time is that they end up with a bad text stack they don’t go back and forth with the large language model in order to actually get the best text stack possible for what they’re trying to do and they end up launching something that is mediocre that makes them run into issues later down the line when it comes to scaling the cost, how fast there is functionality, etc. So before we start building anything, we want to have the text stack dialed. We want to keep going back and forth with Claude until we get the best text stack possible. And then we also want to nail the design so we know exactly what the feel should be of the user experience before we even start building anything. It’s much harder down the line to change something from bad to good rather than just building it from good in the start. After we have done that, after we have spent a lot of time, and I’m literally talking hours here, if you’re building something complex, figuring out the text stack, figuring out the design, and making sure that all of these will be good, now it’s time to build out the implementation plan. What people that are a bit more intermediate, but still beginners, what they do is that they maybe just make one plan of, okay, this is what I want to build, give it to Claude, and then have that build it out. and port should do this next step itself which is building out an implementation plan. But you’re going to get much better results if you actually verify that you have a good implementation plan. An implementation plan is basically a plan for how do you actually build the thing that you have written out in the build plan. And the way I like to do them is split them up in steps. And for each step I want a bunch of different checkboxes. So to give you an example, step one might be to set up file structure. And in the file structure, there’s a bunch of different substeps in setting up the file structure and you want to write all of this out in your implementation plan. And if you’re building something very complex, you might have something like 40 steps in your implementation plan in order to build the thing out that you want. And before you even start building anything, you want to make sure that you have this implementation plan dialed. In step two, the next thing we might want to build out is maybe the database structure and then maybe authentication. Cloud knows the best sequence of how to build out anything, but you want to force it to make a good implementation plan. You want to force the AI to really make a decision and make a good decision for how to build out the thing that you want. And then after you’ve made the build plan and you’ve made a good implementation plan, that is when you start building. And the way you do that now is that you give the large language model this implementation plan of course. And then you tell it, please start with step one. And the large language model is now going to start with this substep, then do this substep, and then do this substep and come back to you and say, well, step one has now been complete. Do you want me to start with step two? And then it’s going to start step two. It’s going to do the substeps inside of step two, and then it’s going to finish. This is much better than just giving it an entire plan or just giving it a build plan and letting it control everything itself. By this, you force it to do it in the best sequence possible. And you have planned everything out before you even start building anything. I hope you start to see the power of building out both a build plan and an implementation plan. And actually, when it comes to like the time that you’re building, step one and two will be where you spend 80% of your time. Step three when building something out. And actually building it will only take around 20% of the time. A lot of people in the AI space get this wrong and they spend 80% of the time just letting the LLM build and only 20% of the time planning it out, if they even do plan it out at all. So when you made a build plan, you made an implementation plan and the AI has gone through every single step all the way down to step 40 or however long the implementation plan is and the AI is not done building it, then it’s finished and you’ll have to test and refine it a lot. You’ll definitely run into bugs. It almost always happens and you basically just go through them, paste the bug to plot and have it finish everything. That is basically how you go about building something. In all of this, you want plot to be almost a consultant that can explain to you what is the best text stack. What should I use for building this? What gives me the best results? And a bonus tip that gives you even better results is if you use two large language models. It could either be two claw code sessions, could also be one claw and then one codeex. But if you have one AI as kind of like the consultant that decides what to use and then you take that decision and you give it to another AI model and you say, “Is this true?” They’re then going to bounce ideas and make sure you get the best result possible. And by doing that, you can start with absolutely no technical ability and you can figure everything out on the fly. Awesome. Let me actually demonstrate exactly how you do this by building something that I have no idea how to build. then I’ll show you exactly how you use this four-step framework right here to do the exact thing that we just talked about. Let’s get into it. To show you that this actually works and to show you how to actually do it, let’s build something that I haven’t built before just to show you that it actually works and that you can build something really, really good without knowing how it works. First, I want to build a lead enricher. What that basically means is that I want to be able to insert a name or an email or a LinkedIn URL and from that I want to search up a database of some kind of sort and from that I want a software that can go out and get all information on that person. So if I wanted to find someone’s email, I could just paste in their LinkedIn URL and that would give them their email. I have a slight idea of how to build this. I can probably use some API from like Apollo or something, but I wanted to show you how you can know none of that and get all the answers from Claude in order to build it out. So, I’m not going to help it and tell it what to do. Claude is going to make the decisions for us. I have a blank project right here. I’m just going to head under the terminal and click new terminal and then open up a Claude session right here and click yes, I trust the author. Then going to drag it on top right here. And I’m going to write I want to build a lead in richer. First I want to build a build plan. Ask me some questions about how it should work. So we can create a build plan as amp file. We just set it to auto mode and hit enter. And now Lord will ask us a couple of questions to basically build the thing that we want to build. And you can actually see it’s even using the superpower skill which kind of forces this entire process. This superpower skill first creates the build plan of what we want to build which is basically what it calls a design spec and then it actually also builds out an implementation plan. This is what’s going on under the hook for the superpower scope. You can see it says what’s the input to the lead enricher? What do you start with? I want to be able to start with anything and from that so either email company domain name plus company linked in URL and from that get all of the information. So, I should be able to insert what I have. Hit enter. What output fields do you want the enricher to return for each lead? Let’s do person basics and contact info and company info. And then, yeah, let’s get socials as well. Submit. Hit enter. Which enrichment data source do you want to use? This is where we want to make sure that we actually make a good decision. Choosing an enricher is the biggest cost of running this. And right here, it’s giving an option to have multiple and use each as fallback. I like that idea. So, let’s do that. What do you want to interact with in Richer? Let’s do a small web app like this. How should leads be input into the web app? Both single bulk one form from a single lead. Let’s just do single lead only for now. Do you want to store enriched leads or is it lookup only? Store history. And if you reach a question where you don’t know the answer, you can literally just scroll down if you’re using the superpowers plugin. You can also just go out of it and then discuss it with Claude. But we can write chat about this and then say what would you recommend? And then it says I’d recommend store history in a database. Reasons saving on cost blah blah blah. The only reason to skip storage is if you have a strict privacy policy compliance reason. Let me just say for now that we don’t need to store it. Now you can see it says got it. Look up only. How should the provider fall back chain work when enriching a lead sequential with merge? Try Apollo first then Hunter. Let’s try and just say let’s just try Hunter first. And I want it to stop when it finds the email cuz that is the main thing. If there’s something that you don’t understand, something that you don’t have an answer to that you think is best, you should just ask cla what it would recommend. Here you can see it says what order should the providers be tried in? Gives us a couple of options. I’m just going to say hunter only, but you can l just go back and forth with claude like this. What’s the text stack? Should I match the shiny automations template? It knows this from memory. Um, let’s do next year only and no trigger.dev for now. I have enough to draft the design. Here’s the proposal, architecture, blah blah blah. That’s all of this. And now it’s creating what’s called the design spec. And the design spec is basically what I called the build plan in my previous explanation of how to build anything. You can see it has two questions, two things I like to decide before I write the spec. How should I handle the LinkedIn URL case? Let’s for now skip URL and let’s do a drop down. The skill I’m using right here, which is also very valuable for you to know about, is called superpowers. And what superpowers basically does is that it forces Claw to think like an engineer. All you have to do is literally just search GitHub superpowers. You’re going to find this GitHub repo right here. It’s called Ora Superpowers. Click on this and you can see it has 28,000 stars. All you do is that you literally just copy this right here and you go to claw and you say, “Please install these skills. search up the documentation for installing global claw skills before you do it. You should just send this like this and it’s going to install the skills for you. After you’ve done that, you might have to restart claw just like this before it’s ready. But now you can see spec written and saved to ducks superpowers and then it has the spec right here and that is sign. There we go. So if we click this, you can now see it build out a full plan for us. And we want to read this carefully to make sure that aligns with exactly what we want to build. So you can see it says lead enrich design spec. It shows the date. It shows the status which is draft a goal a personal web tool for enriching a single lead at a time. The user enters whatever they have email company name or domain name and domain name and company. And the app returns the lead’s contact information company info and social presence. There we go. Non goals. We don’t need a CSV enrichment. This is not what we want. It shows the stack right here. It shows the user flow, shows the architecture, shows the component responsibilities, what each page should do. You can basically see we have a full plan already for exactly what it is we want to build. If we find something inside of here that we don’t like, we can tell claw to fix it in the implementation plan. It even shows how the enrich lead should be structured. It shows error handling. It shows file layout and it shows out of the scope maybe for the future like this. Nice. Now we have our full build plan and right now we’re building a fairly simple tool. It’s just like a personal tool. If you were to build out a full software for example, then this would definitely be an extremely long build plan. You would have to include pricing and authentication and databases and maybe you’re running some virtual machine like hosting the back end. All of this would be inside of your bill plan or what superpowers call it the design spec. And see it says spec written and saved. I’ll say I have read it now. Please continue. Now you can see it’s using a different skill also coming from superpowers that’s called writing plans. And this is where we go to the next step which is building out the implementation plan with these substeps. And building out this implementation plan is actually the thing that usually takes the longest. I’ll let it do that and then I’ll come right back. And there we go. Now you can see the plan is complete and saved inside of plans. And this is what the implementation plan is. Right? If we click on plans and I’ll click on this, you can see we have the entire file structure. We have task one, do this and this and this and this. Set all of these things. Step five, step six. So it’s basically split up in tasks and then steps inside of these tasks exactly like this. So now we have two options. to be one sub aent driven where each thing are built where a fresh sub agent is doing each task. This speeds it up a lot or do we want to do everything in line. I’m going to do one sub aents start the build and only now after we’ve made a spec and after we’ve made a plan is when we start building it out. So now it’s going to build out this little mini app for us which is a lead in richer. And there we have it. Now Claude has built out this entire thing by following the implementation plan and this took 13 minutes. The next thing we want to do is testing and refining. And we are probably going to run into some bugs. You can see we have our environmental values right here. Let me copy that, paste it, and then rename it to just local. And let’s then update this Hansa API key. Go to hun.io and go to API. Copy this API key. Go back. Paste it in right here. Close it down. And let’s try and run the server. So we can now write mpm rundev. That’s going to give us localhost 3000. Paste it in and let’s see if our app works. So it says lead enricher right here. What do you have? Let’s say we have a name and company. And the name we can write this. Let’s test it on myself first. Company name shiny.ai. Let’s see if we can find it using this. Click enrich. Now we run into this issue right here. So I’m going to copy this. This was as expected. Paste it in and say we run into this issue. Hit enter and then we should be able to fix it. It was something to do with how it’s passing the information. So let’s see if it works now. So you can see it’s making a bunch of updates now in the code. Now it wants to test it. We can just stop it here. Let’s try and restart the server. Write name and company email again. Click enrich. And there we go. Now we have a bunch of information. And it could even just from my name and the company name, it could find my email. Let’s see if I just write shiny.ai and click enrich. See what data it then finds. Then you can see it actually finds Oliver, my co-founder, and his email. And there we go. The lead enricher works. It’s not a very impressive app because it literally just uses the Hunter API. But it was also show you this flow right here of building something that you’ve never built before by using these four steps. Creating the build plan, creating the implementation plan, and then having the LLM build for you, but spending a lot of time on the build plan and the implementation plan. And then when you have all of that when you have sharpened your axe, that’s when you start building it. And I guess we can do the next step which is testing it and refining it. We have already done a bit of bit of testing, but let’s refine it as well. I can give it the front end skill now and say make it look clean and modern like a modern SAS. Hit enter. And now it’s basically just going to give our entire software a full makeover of how it looks. That’s a part of the refining step, actually making it look good. There we go. Now we have given it a bit of a makeover. Now, find anyone’s work email. And now we can choose either email, domain, name, company. It works exactly like it did before, but now it just looks a little better. Awesome. The reason I wanted to build this out is just to showcase how you basically build anything. So, when a client comes to you and tells you, I want to build out this and this and this. I want to try and automate this. Is this possible? You can literally just ask Claude about how would I build this? How would I go about this? Use this framework right here for first creating a build plan, creating the implementation R and then having it build out. And that basically allows you to build anything you really want or would never need. If you’re already technical, then you might be able to skip some of these stages. Like you might already say that I want to use this and this and this text stack because you already know that this text stack would be good for the task. But if you’re a beginner and you’re just starting out and you don’t have that much technical expertise, don’t skip the step of just chatting with Claude and finding out what the best tech stack would be for things like scaling, cost, how fast it is, the functionality, etc. Go back and forth with Claude a lot about that and you’re going to get the best results. In the next part of the course, we’re going to talk about pricing because now you know how to create a bunch of stuff. But how do you actually price it? How much can you charge? What pricing model should you use? That is what we’re going to cover now. Now you know how to build a bunch of different things and you also know the framework for basically being able to build anything using clawed code. But how do you actually price your services? How much should you be charging? This is a question I get a lot. And I always use the golden rule that when it comes to developing an AI project, you always want to give your clients a 5x return on investment. So, let’s say that you save your client $5,000 in payroll or in whatever solution that you implemented. You would be able to charge a $1,000 from that for your service. If you keep this 5x return on investment, your client will always be happy. And a happy client means low churn, which basically means that they stay with you for a very long time. If they’re paying you monthly, you of course want them to stay as long as possible. And it increases referrals. If you did a good job for them, then they’re going to refer you to their friends that are also business owners that might also have the issue that you’re solving. And this where you potentially can achieve negative churn, which basically means that your agency grows without you doing anything. Just from serving your clients right now without paying anything else in marketing, your agency would still increase because the churn is so low and clients are referring you to other clients in a higher rate than the churn you have. And that all starts by giving your clients a good return on investment. That’s the entire point of this business model. And in any service business really, you need to go in and do a hell of a job. Give them a return on investment in order to actually deserve the $1,000. But how do you actually calculate the return on investment for your services? It is basic math. Let’s just use an example that you are installing voice agents and and right now you’re working with a client that right now has 10 people in customer support and they’re taking calls all day. They’re answering the same questions over and over again. So now you installed your voice agent system and this now means that they go from 10 customer support reps to let’s say five customer support reps. This means that they’ve cut five customer support reps. Each rep might be paid something like $2,000 a month. So when you have cut from 10 down to five, that is five reps, that basically means that you’re now saving this company $10,000 a month. So using the 5x return on investment rule, that mean that from that, in order to keep the client very happy, you’ll be able to charge $2,000 a month for that. And if you do it like this, then they have no reason to churn. You’re literally saving them still $8,000 every month just from using your system. Of course, you need to do some work. You need to maintain it. You need to make sure that voice agents actually do a good job, but that is also what you’re paid to do. That is how you can calculate a return on investment. You might also have a system that directly generates revenue. Let’s say you have like a cold email AI system that you install for clients and that system is generating a business, let’s say $100,000 a month in new revenue. When you do this, you need to calculate in gross profit because this is a revenue. Gross profit is what you have left after you remove the fixed cost. So things like payroll for taking care of the client. How much is the company actually profiting from that $10,000 worth of revenue? That might be, let’s say, if they have good margins, like $40,000. So that is in theory what you’re making them every single month. You can charge one of that, which means if you’re running this system right here, you’ll be paid $8,000 a month. So, you always want to give them at least a 5x return of their money. And of course, this $100,000 a month in revenue that you’re giving them needs to be new revenue, right? Not their existing revenue already. So, if you can by installing your system, give them a 5x return on investment, then they’re going to be very happy and they’re going to stay with you forever. But what pricing structure do you then use? If you want to make it really easy and if you want to follow the the golden rule when it’s available, when result based, when it makes sense, it’s a very very good offer. If you literally just say to them upfront that you’re going to be taking 20% of the gross profit that you make a company, that is such a no-brainer because there’s no risk in it for them. You remove all of the risk. Now there’s only an upside and it also shows that you’re very confident in what you’re doing. The negatives are of course is that you need to be able to track all of this. So this usually works with larger clients where you can actually go in and control and see okay how much money am I making them that you can follow each lead in the process that they have a good CRM etc. That’s one of the really good offers resultbased and when you are good at what you do resultbased will also usually pay you more than something like a regular retainer. Let me give you an example. Shiny.ai AI, which is our AI agency, does lead reactivations, and we always use result-based. We had one client where we probably could have charged maybe like $1,000 a month on a retainer. But because we did it as resultbased because we went after the golden rule, that client in total actually made us around $240,000 in LTV. So, we probably at least 10x the lifetime value just from doing it result based. Resultbased is one of those offers that you would rather have a retainer if you’re bad at what you do. But if you’re really really good at what you do, you would always rather have resultbased because you’re paid for every result that you bring. So if you can scale that up, your result also scales infinitely. Well, that kind of makes sense. But this offer right here is not always the best one to use because sometimes it could be hard to calculate how much something is worth. Let’s say that you are implementing customer support widgets on websites, right? to this little website widget in the bottom right corner. It’s very hard to say how much a conversation is worth. You could still do result based. It will probably not be of gross profit. Instead, it might be per conversation, for example, that you’re charging them a little something. Maybe you’re charging them something like $1 per conversation. So, this still definitely still add up if they have like 200, 300 conversations a month. But that’s one way that you can do these kind of like support agents as result based. But you cannot really like calculate gross profit directly from like a support agent. So sometimes resultbased really doesn’t make sense. Instead, what I recommend at that point is the upfront offer plus recurring because sometimes it’s simply like impossible to track and it also becomes a headache if you have like 50 clients to track their performance every single month. So charging something up front and then a recurring afterwards is sometimes the best option. But if you do that, I’ll always recommend that you use the satisfaction guarantee. This has worked so well for us. Let’s say that we want to charge a client $2,000 upfront and then let’s say $500 a month. If we just give them this offer right here, there’s a bunch of risk, right? The client is taking a risk. Okay, does this even work? And they’re paying a pretty big amount right up front without even seeing if it’s going to work yet. Instead, what I want you to do is to give a satisfaction guarantee. So, you still charge $2,000 up front. You say, “I’m going to build this out. It’s going to cost $2,000 as a setup fee, but I’m so confident in my own ability that I can build this really well out for you.” So, after I build out the entire system, so if you’re building a dashboard, you will build out everything. If you’re if you’re building a website, you’ll build out everything. You’re basically going to say that I’m going to build out everything so you can see exactly how it looks. then we’re going to hop on a meeting and if you don’t like the result at that moment then you can get a full refund. So then we have the satisfaction guarantee. The reason why this is so strong is because now all of a sudden there’s no risk in it for them. If they don’t like what you have built, then they can just get their money refunded right afterwards. But if you’re actually good at what you’re doing, which you hopefully are, then people are not going to refund because they’re going to see a website that they actually like or they’re going to see a dashboard that actually automates the things that they want automated. It’s very important sometimes for some projects that you get some money up front. The reason for that is that it has happened countless of times for us that we have started the project without taking any money up front and then the client simply like stopped the answering. Maybe he got cold feet, maybe something else came up, maybe he was just too busy. But the reason they could just like stop answering and don’t care anymore is because they were not invested. You want people to be invested right from the start when you work with them. So, this is like the perfect middle ground. You still get them invested, you still get money up front, but you remove the risk by giving them a satisfaction guarantee. And if you’re good at what you’re actually doing, we have given this offer a bunch of times, and we have never had to give a refund. Because every single time we have shown someone a website, for example, or shown someone a system that we have built out for them, they have said, “This looks good. I’m ready to try it. Awesome. Let’s get started.” And for a lot of the time, they actually forget about the satisfaction guarantee when they see something that’s really good. And then I’m a big fan of recurring businesses. I don’t like to work on a project basis. I like having recurring revenue coming in every single month. So when you charge an upfront fee and then charge a lower recurring fee, this lower recurring fee is usually not a problem. And this really helps when you start to stack like 10 plus clients, all of a sudden this recurring revenue is just stacking and compounding. What you basically justify is that this is to keep up support. It’s to keep the system active to fix anything if any issues come up. That is what this fee basically like a service fee covers, right? Which is a monthly recurring fee. Another thing that’s very important to talk about is that you always of course want to make sure that you profit. You don’t want to build out a full system for them and then cover let’s say the software cost yourself and then realize that you that you’re not going to profit. A way to overcome that is to use tiers just like you would with the software where you have like tier one, tier two, tier three that becomes more and more expensive. You can do that as well with your AI services. The key here being that you always want to make sure that you profit. Let’s say that you’re building out the voice agents to use the example from before. Then you could have, let’s say, 100 calls on plan one. Let’s say a,000 calls on plan two and let’s say 5,000 plans on plan three and then have maybe like a fourth plan that’s like 5,000 plus, which is maybe per conversation or something that you can charge for. By doing that, by having different tiers, you always make sure that you are in the green. If you’re just charging like 500 bucks a month and all of a sudden you realize that this company is massive and they’re going to have like 10,000 calls a month, you might get ruined. So, you want to make sure that you’re dialed in on this and that you are always in the profit. A question I get a lot as well is, should I include software in my price? Like, let’s say that you’re building out the voice voice agents as an example. it costs maybe like $100 a month for you to run it. Should you tell that to the client and get them to pay $100 extra or should you include it in your price before? If you’re charging 500 bucks, then it would just charge 600 bucks without telling them. And from my experience, what I’ve seen is that a confused person doesn’t buy. If someone is confused about what they’re going to pay, what they’re going to get, they’re not going to purchase anything. So, you want to make your offer as simple as possible. Instead of saying, “Well, it’s like $500 a month for the service and then $100 a month for the actual software,” we almost always include it all in our price and just have a very simple offer that is easy to understand. There’s one scenario, however, where we don’t do this, but where we actually do this, and this is if the client wants to like own the system completely or like run it on their own servers. This is usually like bigger clients that wants to run everything on their own servers. The price tag for that is of course going to be much higher as we now need to install it on their servers. And when we’re doing that, they are of course also paying for like the hosting and the software cost. So I’d say when you’re just starting out like 99% of the time you would just include the cost inside of your price. When you start going up market and working with larger clients, that’s sometimes when you have to sell them. Well, it’s going to cost this and this on your own service and you basically take the price out of your service fee. I mentioned it a bit earlier, but please, this is just like a reminder. You never say the price of the service before you have shown the value. It’s what we talked about, right? You don’t, for example, if you’re talking with someone in the DMs over email and they’re going to ask how much is it going to cost. You never say the price before you show what you’re actually delivering. You also want to say on the call that you’re trying to give them a 5x return on investment. They need to see the value that they are going to get to compare it to the price that they’re going to pay. You want to be judged on the value, not on the price. And the only way that you can do that is that you hop on a call with them. You talk about the issues that they currently have and then show them how your solution could be a solution to their problem. If you [ __ ] this up, then you’re never going to get them on a call because they’re just going to see, well, it’s $2,000. That’s way too much in debt. But if they realize that it’s actually going to save them $10,000, then paying that $2,000 is a steal. And the other thing is that a lot of people when they start selling in the AI space, they are selling AI. They’re f focusing on the features like we are using the newest models. We’re using the newest technology. We are running it smooth so it’s as fast as possible. The reality is that most companies that you’re going to work with doesn’t really care that it’s AI at all. All they care about is the outcome that they’re going to get. Will they be able to save $10,000 a month? Yes or no. They don’t care about if it’s AI or what it is doing it. They just care about the outcome. So when you are on calls and you’re explaining the value that you focus on the pain that they have right now, the amount that they’re paying extra right now, which they don’t have to, and then from that show them the outcome. Show them what you can help with. Don’t focus on the features. Don’t focus on it being AI. I hope that clarified some things in terms of pricing. Now you have all of the information that you really need in order to go out and actually start selling AI services. But in what sequence do you actually do it and what does your road map look like? Now that is what we’re going to talk about in the next part of the course. Now you have literally everything you need in order to go out and land your first clients. Don’t let this just be another YouTube video that you watch, forget about, and where you never actually take action. Go out and actually use this information. To make it extremely easy for you, let me give you a road map that you can do right now. The first thing is what you’ve actually already completed. it is to learn a skill. Just from watching this video alone and understanding how to build with tools like cloud code, you already now know a valuable skill that you can go out and sell. You should of course still keep learning as much as humanly possible. The AI space is always moving. So you need to stay up to date, but don’t get in the trap of overarning and never actually going out and utilizing this information that you’re learning to anything. So, while you’re still learning and while you’re still keeping up to date with the AI space, now it’s time to go out and land clients. And these clients are not going to be the one that pays you the most. Start by working for free. Reach out to people on cold email like I’ve shown you how to do this in course and offer to design a website for them for free. become their internet partner or get started on Upwork like you also know how to do now and charge something low like $15 an hour just to get your first project done and to build that Upwork profile that can become a very very valuable asset. One of the biggest mistakes beginners make in this space is that they think well I’m starting an AI agency so I should spend most of my time actually learning AI but that is actually not the truth. The main thing you should be learning is how do you market it? So, how do you do marketing for your services and how do you sell? Those are the first things that you need to figure out and those are usually the things that are the hardest for most people in this space. People massively underestimate how much time they should actually be spending on marketing and selling. If you’re just starting a business, try and spend 4 hours every single day doing marketing and doing selling. So spend four hours outreaching to everyone you know, sending out Upwork proposals, building code email campaigns. Do everything you can in order to land that first client. And again, the first step is not going to be to make a bunch of money. You’re still building the skill, which is the actual valuable part. The next step after landing a couple of projects and landing a couple of clients is literally just to sustain yourself. And when you talk about making it a business, right, the success criteria of making it in business is to stay in business. This is the point that you need to get to because if you can just sustain yourself, if you can just pay yourself a small salary every single month that covers your rent so you don’t have to worry about it, that literally means that you can go all in and that you will never give up. People say that nine out of 10 businesses fail, which is true, but the majority fail right here at step one and two. And that’s because they never reach the point. So at one point they say, “Fuck it. Let me try something else.” And they give up. Or they might just be false to, well, this month I can’t pay in. I need to go out and get a job, so I’m going to quit with this or hold off for a bit and then never get started again. See if you can get to this point right here. This should be your main goal right now. Can you just sustain yourself at $2 to $3,000 a month? Get to that point because then you never quit again. And then after that, you reach step four, which is that you start charging more. At this point, you should have a consistent stream of clients coming in either through Upwork, either through cold email. And when you have that, that’s when you charge more. You choose the clients that can pay you more. when you say no to the irritating clients that are broke and that can’t pay. So, you work with fewer clients that pay you more. Step number five, now you need to find bottlenecks. Scaling a business is basically just the process of finding out the bottleneck, solving that bottleneck, and then going to the next bottleneck. Let’s say that this is a funnel that looks like this. Let’s say that we’re getting a bunch of leads from cold email, right? They all come through here. But then we have a sales process right now where it’s only us taking sales calls and this only means that we have time for some calls and our calendar’s completely booked up. The bottleneck of this business right here is to hire more salespeople or by figuring out how can we only get the most qualified through this funnel. So maybe you open this up and all of a sudden you get way more calls and way more closes. Now the bottleneck might be that you only have yourself as a tech person. So then you need to hire more tech people or you need to automate more with cloud in order to make the onboarding process more smooth, better, etc. But that is the process of scaling a business and that’s step five. And honestly, you’re going to be at step five forever. That is the process. Now you’re just finding bottlenecks, solving those bottlenecks and this basically just keeps going on forever as your business grows. This course is about building a one person business with clawed code. But this doesn’t mean that you should limit yourself to just being one person. You start out by being one person in the business and Claude code can definitely carry a lot of that weight and doing it and do a lot of work for you and you can probably get to 10 15 clients yourself where Claude does most of the work. However, don’t get stuck in this mindset of okay, I shouldn’t hire anyone because sometimes the best solution is not AI and that’s even coming from me. Sometimes the best solution is to just find a good salesperson that can convert more calls, you can close more deals. Or sometimes the best solution is literally just to find another tech person that can onboard more clients and make you more money. A mistake that I see a lot of people make is that they try to overmate where the easy solution that they could have done in a couple of days is usually just hiring someone and instead they spend weeks building out and automating a solution that then turns out mediocre. But that is the endgame. The reason why this course is built a oneperson business is because that’s how you start and that’s also how you can get to make a really good living for yourself just working by yourself with cloud code. I hope that this road map is valuable for you. Write it down and literally just start outreaching. Start setting up your cold email campaigns. Start setting up work. Buy connections. Send proposal requests. Actually spend a bunch of time doing these Loom videos and just keep going. When I started my business, I was considering myself smart. But it still took me 4 months of just straight grind outreaching every single day before I landed my first client that paid me 400 bucks. These expectations of being able to make tens of thousands of dollars your first month is just completely unrealistic because you haven’t learned that skill that makes you worth that much. In order to do that, you actually need to build something up that can take months if not years. And the best thing that you can do right now is to set your expectations right from the start. Say to yourself that it’s okay if I put my heart and soul into this and that it takes one year. Don’t expect it to take a month. And when it comes to growth in a business, it’s usually exponential. You’re going to have a pain in the start. You’re going to be hovering around a couple hundred bucks a month probably when you’re just starting out. But then all of a sudden, something clicks and that’s when the graph goes like this. And just like the stock market, it doesn’t go in a nice curve. It goes like this, right? It goes up and down, up and down. And then you’re going to have good months. And then you’re going to have terrible months. And you’re going to have even better months. Then you’re going to have even bigger falls and then it goes like this, right? That is how it always goes. Take it from me. Keep grinding and it’s going to be so worth it when it actually clicks. The last thing I want to say is that if you’re already in a job or already in a position right now, don’t let that stop you. It’s a myth that you have to quit everything that you’re doing right now in order to go all in on something. Because for a lot of people, that actually keeps them from starting in the first place. Because what if you cannot afford to quit your job? That is a big risk if you don’t make this work the first couple of months. Instead, what you do, and this is by no means easy, is that you build your business from 5 to 9. Then you are at work from, let’s say, 9:00 to 5:00, and then you might be able to get a couple of hours in at the evening before you then go to sleep, right? That still gives you 4 hours right here and 2 hours right here. And usually you’re the most aware right when you wake up. So spend this time working on your business. And then probably in weekends, you probably have 10 hours that you can spend on your business. So this you have to go all in when you’re just starting out. It’s a myth. You can start getting your first couple of clients, making your first money, setting up your Upwork profile, and do all of that while you’re still working a job. Don’t quit your job. Don’t think that you need to quit your job. Do it on the side until you can actually afford to quit. I really appreciate that you watch this full free course, and I really hope that you found it valuable. If you want my help, then you should check out our community, the 1% in AI. Inside of here, we have the 30-day challenge and the 90-day challenge. And the great thing about these challenges is that when you complete them, then you actually get your first month completely refunded. The way it works is that you get this tracker right here that shows you what to do every single day, what modules to watch inside of the 1%, what task to complete, and it basically holds your hand throughout the entire process of learning the skills that you need to learn and setting up your AI agency. So, if that sounds interesting, then I’ll leave it as the top link right below this video. Thank you guys so much for watching. I hope you have a wonderful rest of your day. I hope that you’re actually going to use this for something. Don’t let this be another YouTube video that you’ve just watched and never actually took action. I am rooting for you.

Imagine it’s a few years from now. AI can build almost anything, write almost anything, and do most of the tasks people used to get paid for. In that world, I got to ask, what skill is still valuable? I’ve been thinking about this non-stop, and I’ve narrowed it down to six core skills. None of them require a fancy degree or connections, and all of them could be started this weekend. And every single one of them gets more valuable as AI gets better, not less. By the end of this episode, you’ll learn the six most valuable skill sets on the planet with some tips to get started. I’m going to go through each of them super clearly so you will understand where the world is going, what you need to learn, and just concretely how to get started. Let’s not waste any time and let’s go right through it.

baby. So, the first skill that I think is the most valuable skill right now is people who can set up agents properly, manage them, and run local AI models. This is basically the grown-up version of prompt engineering. So, a lot of people learned how to type a good prompt into chat GPT, which is useful. But the next layer is being able to design a little AI employee that has context, that has tools, that has permissions, that has memory, that has a goal, and a way to check its own work before it bothers you. That skill, that little skill is going to be valuable because most companies are about to have the exact same problem. They’re going to have 10 AI tools, 50 workflows, a bunch of halfworking automations, and nobody understands how to turn that into an operating system, which is what they really want. The person who can walk in and say, “Hey, here’s a customer support agent. Here’s a research agent. Here’s the sales follow-up agent. Here all are all the rules. Here’s what it’s allowed to do. Here’s where it needs approval. Here’s how we know if it’s working.” that person becomes really really hard to replace. And the local AI part, which I’ve talked about on this channel before, I’ve done a couple videos on it, is important because there are certain workflows where privacy or cost, you know, the prices of these models are going up and up. Latency or control matter a lot. If you could run these uh models locally with something like Olama, LM Studio, you start to understand what can happen on your own machine and what needs the cloud and what needs to touch private docs and what should stay behind the wall and how this all interacts. So I think this whole idea around local is going to be more and more important as time goes on. Even if because a lot of people you know in the comments will say well local models are going to going smaller. Even if local models go smaller, you learn the architecture for the future. You learn which jobs need a giant brain and which jobs just need a reliable worker that never sleeps. So, what would I do? What’s a rep I would do to learn this skill set? The first rep I would do is simple. Build a daily briefing agent for yourself. Give it three sources. Give it your calendar. Give it a folder of notes and give it a a few saved links and its job is to tell you what matters today, what decisions are waiting for you and what follow-ups you owe people. Then you can add one rule. You can say it has to show sources and ask for approval before sending anything. That one project teaches you context. It teaches you retrieval. It teaches you tool use. It teaches you permissions and it teaches you eval. So it might sound small and maybe a boring first agent to build. But I I believe that that is basically the shape of every serious agent inside a company. The mistake people I think make is they try to build an all- knowing agent first. These really big agent pro uh projects. Uh the better move is basically just to build a small agent to start, get it to be very valuable, schedule it, have a clear success metric. Um and you know the success metric might be did it save me 10 minutes? Did it catch something I would have missed? Did it produce something I would have actually used? Um if the answer is is yes, then you’re learning the skill. So that’s the first super super valuable skill. And by the way, this is in no particular order. These are just the six that you know, the more you know, the better. The second skill is marketers who know how to build distribution. I think this one is underrated because people confuse distribution with posting. Uh distribution is way more deeper than just like posting on social media. It’s knowing where attention already lives, what people are already anxious about, what language they use when they describe the problem, and how to turn that into trust before you ask them to buy anything. In an AI world, we all know building products are, you know, is really easy. Uh, building demand is just getting more and more important. So when anyone can ship a landing page or an app, you know, or build a SAS, uh the bottle the bottleneck moves to the question of can you make people care and that’s where someone who’s really good at distribution uh is a pro at. So the marketers who are going to win in this agentic era are going to be part researcher, they’re going to be part storyteller, they’re going to be part media operator, and they’re going to be a part community builder. They’re basically going to know how to take one insight and turn it into a tweet, uh a short form video, uh a YouTube title, a newsletter angle, uh a landing page headline, a founder story, and a sales conversation. So, in a sense, it’s almost like a marketer is becoming like a general a generalist marketer. We’re seeing this as a as a greater trend like people are becoming generalists. they can do multiple things because if the if their job is to manage agents uh they they need to understand different components of that and that’s exactly what’s happening in the distribution marketing uh world. So the first rep I would do if I’m trying to learn distribution uh is I would do a distribution map. So, I pick a niche I care about. Uh, like dentists using AI or solo consultants, maybe real estate agents, Shopify operators, whatever could be a business that you want to start or you you know you’re you’ve you’re building already. Then write down the 20 places the attention, their attention uh goes. Uh, so like the newsletters, the creators they pay attention to. Maybe it’s like Reddit threads that get popular, Slack groups, podcasts, events, search terms, the tools they already pay for. After that, write one painful sentence they would actually say out loud. Something like, um, I know I should follow up with leads faster, but by the time I sit down and do it, half of them are cold. That sentence is where distribution starts because you’re getting you’re basically like transporting or or yourself into their shoes. Um then you should do the second rep. So the evolution of this write 20 hooks for the same idea. Make some curiosity hooks, some fear hooks, some status hooks, some money hooks, some I wish I knew this earlier hooks. If you want to become great at distribution, you don’t you not you don’t want to ask yourself, how do I promote this after the product is already done, right? So, you want to start asking yourself what existing desire am I pointing this at before you build. And I think that that shift alone, that mindset uh is going to change the quality uh of your ideas. And that’s what makes someone who’s really good at, you know, distribution in this era. So, you know, TLDDR uh on on distribution is you want to put yourself in their shoes. Um and you want to be this like part storyteller, part um uh researcher, part media operator and and really just have a lot of uh shots on net in this in this world because, you know, some some are going to win and some aren’t going to win. Um, and on this on the Startup Ideas podcast on this channel, um, you know, I’ll share more of these tools as as I’m learning in real time. So, you know, feel free to like, comment, and subscribe to to get more of this in your feed. The third most valuable skill is robotics engineers who can basically build hardware, wire in AI, and source manufacturing. Now I know that most people probably a very small percentage of people actually has any uh experience in robotics engineers. The reason why I put this in here um before I explain exactly how to do it is because you know software was an incredible business uh for the last 20 years building SAS and there’s still opportunities in in building SAS building consumer mobile apps enterprise apps but the mode is is moving to hardware and I’ll explain more uh I’ll explain more about this. So, um, and by the way, I think a lot of people are sleeping on this. Um, basically the last decade, the internet rewarded people who move pixels around. Uh, I was one of those people. Um, but the next decade is going to reward people who can move atoms around, too. That was like my big insight. Um, robotics used to be this PhD feeling thing. Expensive parts, custom hardwares, weird tooling, long timelines. you had to go to school, get a PhD in robotics to go and build something like that. But now the world we live in is is a lot different. You have this opensource robot learning projects. You have cheap cameras, you have lowcost arms, you have better simulation, you have multimodel uh moto models and you have community sharing uh data sets. So you have you know companies like hugging face who has low robot which is basically trying to make robot learning more accessible and even on hugging face which is like a a almost like the way I think about it is like a database of all these open- source projects that you can go and download those repos you can find something some some really interesting open source technology and inject it into a robot uh and who knows that could be the next big thing the next big Um there’s lowcost ARM projects like SO 100 and the SO101 ecosystem and there’s also smaller uh vision language action models like the uh small VLA that are pushing towards robot policies you can train and run without needing some uh giant industrial setup. So the interesting thing here is it goes beyond the AI layer. Uh it’s the person who can make the whole loop work, right? So, it can you get a cheap arm on your desk? Can you mount the camera? Can you collect demonstrations? Can you train or fine-tune the model? Can you make the robot repeat one useful task? Can you look at a supplier listing in China and understand if this thing’s actually manufacturable? So, the person that can do all three, the building, the hardware, wiring, and the AI, sourcing, manufacturing, understanding that, come on, that is just some skill set to to know. And this is something that I’m learning uh in real time um because I think it’s so important. If I wanted to learn this skill, uh my first web uh rep would be extremely concrete. So I would buy uh or assemble a lowcost robot arm. Um you can add a cheap camera to that. Uh and then I would teach it one boring task like sorting three objects. uh pressing a button or moving it from one tray to another. Then I would document every failure. The camera angle was was bad maybe or the lighting changed. Um maybe the gripper slipped or the data set was too small. Um the model looked smart in one setup and then fell apart when the object moved like 2 in. That’s kind of the point. This is how you learn a skill. Um, robotics specifically teaches humility, humility, uh, pretty quickly. Um, and that’s the humility that becomes your expertise. On the sourcing side, this is really important. I would learn the basics of working with suppliers. A lot of us listening to uh, this podcast, you know, we’re we’re software people. We like digital products. Um, but it’s also important to learn physical products, too, because of all the opportunities that are coming. So go on Alibaba or a similar u marketplace. I’m not affiliated with Alibaba. Um and study how components are sold. Ask for a sample before you talk about bulk. Um you can ask for motor specs. You can ask for controller board details. Uh you can ask for CAD files if they exist. Uh replacement parts, lead times, minimum order quantity, shipping terms, and a short video of the part doing the exact thing that you need. So you’re you’re going to be learning a new language. Um and the language is you know can this actually be made, shipped, repaired uh and used by a normal person. Um this skill is so rare uh because it sits between worlds. Uh software people often avoid hardware um and hardware people sometimes avoid distribution and AI. So, the person who could connect open-source AI models, physical prototyping, and manufacturing has a shot at building things that feel like science fiction but sell like practical tools. And I think there’s just so much opportunity here. The fourth undeniable skill to learn right now is cur curators. So, understanding how to curate who are good at yapping and can do short form video in their sleep. I’m sure you’ve seen these people uh you know yapping uh on Instagram and they’re taking over your algorithm and I mean yapping in in the best possible way. you know the internet is drowning in information and the person who can make sense uh of it in public um is you know very valuable and curation has evolved past past like here’s five links in a newsletter right curation is like uh here’s you know five products that I think in this niche that you’d really like and explained in a really storytelling really cool way so the curator of the agentic era watches the timeline and says this matters because dot dot dot like they understand that they can see a new model demo or a weird startup launch, a robotics clip, a policy change, a news item, a pricing update, a story about XYZ, and they can translate it for that particular niche. What should you learn? What should you ignore? What should you try this weekend? What is hype? What is actually useful? So to be amazing at content, this is sort of the big insight. Uh, which might sound obvious, but the the the insight is like, you know, you don’t need to be super smart to get millions of followers in social media or or just doesn’t even be millions. Get 50,000, 100,000 followers in a niche and build an incredible business around that. Um, by creating net new content, you can just look at what’s happening in your niche and curate really interesting things in short form in an authentic way where you’re just yapping to your phone. And the algorithms right now are prioritizing yapping. Why are they prioritizing yapping? They are prioritizing it because they are seeing AI slot move into the timeline and people are getting tired of that. People don’t want that, right? So, there’s nothing more raw than authentic and being like, “Hey, my name is Greg Eisenberg and I suffered from XYZ um until I found these like five really interesting products or this story or I met this person that helped me and let me tell you about it.” That is the type of content that the algorithms and the timeline are promoting. So before I give you the rep, actually I would say like the yapping matters so much. Like learning how to yap is just is is just such a good skill. Um and the people who are great at it frankly uh make you feel like you’re getting in this like group chat of a research report because it’s fast, it’s opinionated, it’s useful and a little entertaining. um you know those people in your niche that you see them yapping and you’re like, “Oh my god, they’re like I feel like I know this person, right?” So, let’s say you wanted to actually, you know, become a curator/Yapper or learn about it. I would do uh a 7-day curation sprint. So, for one week, pick a lane. Um maybe it’s AI agents for real estate, maybe it’s robotics for small businesses, you know, whatever niche it is. But every day find three things and make one short video using the same structure. I saw this. Most people will think it means this. I think it actually means this. Here’s the move. That structure forces you to have a take, which is the difference between curation and forwarding links. The thing with a lot of yappers and curators that do what they do really well is they have a take. They’re either anti- something or pro something. So, have a take, right? Um, the key thing here is you’re going to want to build uh, you know, some people call a swipe file, some people call it a taste file. Basically, a document of examples you love, great hooks, great analogies, great titles, weird use cases, comments that reveal what people are like genuinely confused about. Curators are obviously only as good as their taste inputs. So if your inputs are generic, your outputs are going to be generic. Um if your inputs are weird and specific and high signal, people are start are going to start coming to you and trusting you because you consistently find the thing before they do. So it’s worth you’re worth the follow. Um so that’s an incredible skill set to learn. Um the next skill set, uh the fifth skill set is what I call the building distributor. So it’s the person who can ship both the product and get in front of people. So this might be the most important skill set if you’re a founder if you want to build a business. Um because for years uh there was this clean split. Uh one person would build and one person would sell. You know you’d have your Wniaak who was like the technical person and then you’d have your Steve Jobs who was like the marketer salesperson. You know, one person writes code, one person writes copy, one person makes the thing, one person gets the attention and AI in this agentic world is compressing this split. So one person now could prototype the product, make the landing page, you know, write the launch thread, you know, make a video about it, record the demo, DM the first 100 users, edit short form clips, iterate based on feedback, pretty, you know, pretty much everything. Um, so that person has leverage because they don’t have to wait for the handoff. They can complete the loop themselves. So nothing really gets lo uh lost. So the loop is the whole game because if you build something small and you put it in front of people, uh, watch where they get confused, you know, change the product, change the story, try again. Most people only do half. They build it in private forever or they talk about it in public forever and they actually never ship the thing. the builder distri distributor uh learns by cycling between both. I think um this is so cool because I think that you know when people talk about the one person$1 billion dollar startup you know Sam Alman talks about it um you know I think that person is going to be the builder distributor and when you start seeing people like Peter who who founded openclaw you can tell that he’s not only incredible at building like he built an incredible thing with openclaw technically but he also is you know if you go to his ax and you just see how he speaks He’s an incredible marketer, too. And and he’s incredible at, you know, doing customer support and so many things just by little things, you know, I don’t know him personally, but just little things that I’ve picked up on. This is one of those people, right? Um so, let’s say you want to be um good at this skill. Well, you know, what’s what’s a one quick thing that you can start by honing in on your skill? So, I would do what’s called a 48 hour loop. So pick one tiny problem you personally understand and then build the smallest version with AI. Um it can be ugly. Uh it can be a script, a form, a simple web app, an automation, you know, anything. Then create 10 pieces of distribution for it before you even feel ready. So that could be one demo video, three short clips, you know, maybe three posts, two DMs, the people have the problem, and then a landing page. You’re basically training yourself to stop se separating the product from the market. So what’s powerful is that AI makes the building part faster. We all know this. So the marketing part uh learning can start way earlier. Uh you don’t need to spend now 6 months wondering if people want it. You spend a weekend building enough to earn a real reaction. And then the builder distributor is dangerous because now all of a sudden you can turn attention into product feedback and product feedback into better attention. So it’s this beautiful loop that you can end up uh be building. Um and the only way to become an an incredible builder distributor is you got to spend more time building and then spend more time launching and distributing and then building that loop. Um I talk a lot about on this channel this concept of the ACP funnel. It’s how it’s the future of building businesses. You know, you build audience at the top, then you convert that to conu community and then you build a product there. And that’s also a loop that uh you know, the builder distributor uh is excellent at. The last skill that I think is just so valuable in this era is IRL community builders. So, this one feels almost old school, especially, you know, because we talked about like AI, robotics, and open source technology and local AI, but that’s kind of like that’s kind of why I like it. Um, as more work moves to agents and chats and tools and feeds, real rooms actually become more valuable. People still want to meet other ambitious people. Uh, they still want to meet people like them, uh, who are into the same things that they might be into. Um, and they also still want trust and they want energy and they want to be around others who who teach them things or who are they just entertained by. So AI makes content abundant. Um, it makes software abundant. It makes advice abundant. Um, so where does scarcity move towards? Well, scarcity moves towards belonging, trust, and context. Who do you actually know? Who would answer your text? who would help you hire, who would you intro, who would introduce you to a customer, who would tell you the honest version of what’s happening in their market. The IRL community builder knows how to create that. They know how to pick the right room, set the right topic, invite the right mix of people, and create a ritual that people want to come back to. A great community is actually more like a habit than an event. same time, same kind of people, same promise, better conversations each time. The first uh rep, if I was trying to get good at IRL events, which by the way, I should I should note like there’s hundreds of millions or, you know, billions of dollars up for grades for people who can create incredible events. And I’m not just like pulling numbers. Well, you know, just like randomly like you can look at huge event companies. um that just absolutely crush it. Um selling events like you know if you even look at tech like look at um Saster uh I think Jason Lmin created an event all around SAS um and you know I think he’s publicly shared some of his numbers like they’re massive. If you look at um South by Southwest, you know, these are huge events. Um you know, these are there’s just a lot of opportunity. I think and I think like in general people don’t want massive events anymore. They actually want these kind of smaller, more bespoke events and that’s where the opportunity lies. Um okay, so let’s say you wanted to get good at being the IRL community builder person. you know what’s one little thing that you can do to learn the skill? Well, why not why not host, you know, six, seven, eight people around one sharp question. So, you don’t start with this massive event. You start with like a dinner or a walk, a hike, a breakfast. And the question could be, you know, if you it could be something like what skill are you learning because of AI? Let’s say if you wanted to build a community around AI, it would be something like that. or what are you automating in your company right now or what do you think everything in tech is is missing and then you invite people who can actually answer it. Then after the event you send a short recap with the best quotes, inside jokes, maybe ideas and what one followup everyone should do. The recap is important because it turns the room into a network. So that’s the whole goal with the whole IRL community builder. How do you turn these rooms into a network? because that’s what creates memory and it gives people a reason to forward it. It makes the next invite easier. So over time the room becomes a media asset, a recruiting asset, a deal flow asset, and honestly a life asset. So I think this skill uh pairs beautifully with others. The agent person who can build tools for the community. The marketer can grow it. The curator can turn the best conversations into content. the builder distributor can launch products from it. The robotics person can bring the weird demos and that’s when it starts getting really interesting. So, um there you have it. Those are the six skills that I think matter most in the Agentic uh era. So, TLDDR, what are the six skills? One, people who can set up agents properly, manage them, and run local models. Two, marketers who know how to build distribution. Three, robotic engineers who can build hardware, wire in AI and source manufacturing. Four, curators who are good at yapping and can do short form video in their in their sleep. Five, the builder distributor, the one person who can both ship the product and get in front of people. and six, the IRL community people uh the the IRL community builder who’s bringing people into these rooms and starting networks from an IRL uh perspective. The bigger point of all this is that the future favors the person who can combine all these capabilities. Um there’s obviously too many tools to know all of them and the advantage goes to the people who know how the pieces fit together. Can you make agents uh useful? Can you get attention? Can you build physical things? Can you explain what matters? Can you ship and distribute? Can you bring together in real life? That is that skill stack. And this is the six skill sacks that you know, pick one and you know, you don’t have to be amazing at all six, but pick one and get dangerous. You know, pick two and you have some leverage. Um, but pick three and you become, you know, the kind of person that everyone wants on the team, in the room, or building the company. um we’re in for a crazy next 5, 10, 15 years about what’s going to happen uh in the job market, in the economy. No one really really knows. But the one thing we do know is that knowing these skills or or skills in general, it’s going to make you it is your defense. It is your shield. You know, I believe that. So, I wanted to make this episode because um I think a lot of people like know that they should be doing something, but they’re not sure what they should be doing. So, I wanted to put just this into one place that’s really simple, easy to understand, the six most important skills. You know, share this with a friend who, you know, might might this might be helpful to uh hope it hope it has been helpful to you and I’ll see you at the next one. Thank you so much and have a creative

Hermes Agent is officially the best AI agent ever made. They just released an update that has several new features that completely change how I use Hermes Agent. In this video, I’ll cover the seven new features you need to implement into your workflow today, plus towards the end, show you how you should be using Fable 5 with your Hermes Agent. This is a mustwatch. This is a banger. Now, let’s lock in and get into it. All right, I’m starting off with one of the best features ever implemented into Hermes agent and that is mixture of agents. What is mixture of agents? Well, let me show you. When you type slash MA and then you put in a prompt after it is going to initiate the mixture of agents. Now, what is mixture of agents? Well, what’s going to happen is my aggregator model, which is my orchestrator, is going to take that prompt. So, that’s whatever the model you’re currently using. For me, that’s Opus 48. It’s going to take that prompt and it’s going to send it to a bunch of reference models, right? So, we’re taking this prompt. It’s setting it to a bunch of reference models. These reference models by default are chat GBT as well as deepseek v4 through open router. That’s the kind of the default way they do it. It takes the prompt, gives it to all these reference models. The reference models answer the question and then give the response back to the aggregator model which again opus 48. The aggregator model takes all their responses and then synthesizes it into one response. One synthesized answer according to new researches benchmarks. This results in a much better higher quality response. They tested all the models. They tested a bunch of different reference models. their best combination of models was again Chad GBT and DeepSeek and that is kind of the default in here. All right, so all you need to do to make the mixture of agents works is click on the dropown, choose MOA default. That’s going to send it both to chat GPT as well as DeepSeek 4 through open router. I’m going to give it a prompt based on what you know about me and my interests, passions, goals, and ambitions. What are five SAS ideas I can build out today to make money online? That’s a prompt you can steal right now. Feel free to steal that. Use that with your mixture of agents. Hit enter. Now all the agents are going to get together, convene, chat with each other, figure out what are the best five SAS ideas based on all the conversations, all the memories it has about me and give me a really nice synthesized list of ideas I can start building out immediately. So if I want to turn around after this, go to Hermes and say, “Hey, build that idea out. I really like that a lot.” It can do that. So Chad GBT is talking. Deep See, Deep Seat’s talking. Chad GBT. They’re coming up with different ideas. And here’s a nice synthesized answer right here. And boom, here’s an actually a bunch of awesome really, really good ideas for SAS. I can take one, start building out now, and boom, I have a weekend project for myself. Do the same thing. Set up mixture of agents. If it doesn’t work for you off the rip, it’s probably because you do not have open router set up. Open router is default. The second model uses the default first models, chat GBT. If it’s not working, make sure you have those API keys in. Ask Hermes. It’ll set up for you. Really, really easy to do. mixture of agents. Awesome new feature to get even better answers out of your Hermes agent. The way I’d use this is just whenever you have kind of more complex questions that can go many different ways. I use the mixture of agents for it. Makes it really nice to use. The next major feature in this release is slasharn. This is an amazing way to teach your agent new skills. Basically, the way it works is you can hand your agent any URL, any prompt or run slasharn after any task, and it will automatically create new skills and memories around what you gave it. Let me give you a good example here. So, if you’re anything like me, you’re on X all day. You see tweets with cool tips on how to use your agent, how to use Claude code or whatever, and oh, I’m going to go recreate that in my Hermes. Now, what you can do is take those tweets and just do slasharn and have your Hermes make a skill based on every tweet you see that’s helpful. I got this tweet here by me. Great tweet on coding loops. I’m just going to take the URL for this tweet. I’m going to go to my Hermes agent. I’m going to do new session. I’m going to do slasharn paste in the URL and I’m going to hit enter. And what’s going to happen is the agent’s going to go read the tweet and then turn it into a skill. Now instead of having to explain every single thing you see online, trying to make skills out of it manually for your agent, you just give it a URL, do slasharn, and boom, you have a brand new skill automatically made. Slasharn is the quickest, easiest way to turn anything into new skills automatically. Anytime you think of a new process, anytime you see a helpful tweet or article online, you just do slasharn and then type it in after. You can even do it where if you have your agent do a whole long process, you’re doing something, you have it build something, you’re like, “That worked out great. I want that to be a repeatable skill, you just do slasharn and hit enter, and it’ll take everything from before that and turn it into a skill.” It is a really good way to quickly self-improve your age. And as you can see here, boom, new skill just made based on that one tweet. Every time I see an interesting tweet, I do slash learn, put the tweet in after, and it’s good to go. Slasharn, use the hell out of it. This next one is an awesome one that builds off of slasharn, and it is slashjourney. This is a great way to see everything your agent has learned and all the skills and memories it has. Look at this awesome chart. Now you can see every memory, every skill made from all of your conversations you’ve ever had with your Hermes agent. And you can see how they’re all connected. You can see how all the skills connect to all the different memories. I started building a game. It showed me all the memories attached to building games. Here’s my work around building local models and putting them on my computer. You can see all the memories around local models. Really sick way to see what your agent has learned about you as you’ve worked with it slashjourney. You can see this amazing chart and just a good way to make sure your agent is learning along the way and all the self-improvement it’s had. Which brings us to the fourth improvement of this new update, which is self-improvement got cheaper. Now, every time your agent learns something, every time you do slasharn, every time it self-improves automatically, it actually automatically switches to cheaper models to implement those learnings. A big part of the cost with Hermes aging is there’s a lot of complaints about cost, especially if you’re using anthropic models is it’s really expensive. Well, the issue is is a lot of the things that happen in the background. Creating memories, creating skills, learning new things, those are very expensive tasks and it uses your Opus model, if that’s what you’re using, your Chad GBT model to do those things. Now, it’s actually delegating those tasks to cheaper models in order to save a lot of money. So, if you’re using Hermes Agent regularly, you should start seeing much better cost savings as you go along. Major update, really awesome to see. Next up is massive improvements to vibe coding. Now, when you vibe code inside Hermes desktop, you have full git controls. You can see code. You can see diffs. You can make commits. You can open up pull requests. It’s all built into Hermes. Now, this is an actual viable vibe coding tool. If you look here, I asked to build a 3GS game, which for the record, I’m going to show you the game in a second. I actually built is incredible. Probably the best 3GS game I’ve ever seen built out of this. But what you can see here is it creates all the files. You can see all the diffs. You can see all the code changes. And if we go over here to the right, you can actually make commits and actually open up pull requests straight from Hermes. So, this is a viable vibe coding tool if you want. You don’t need to use claw code anymore. You can use this and just build out your code and your apps straight from Hermes. And Hermes is a very good vibe coding tool. Check this out. I just asked to build an action RPG like Diablo and this is what it came up with. This is actually really, really nice. You can see the graphics are good. You can see the hit counters. You can see the spells it gave me. It has full levels. It has leveling up. This is really amazing. This is a really good game that Hermes vibecoded all by itself. This is probably the most impressive uh vibecoded 3GS game I’ve ever seen. This was Opus 48. I’m getting really into Path of Exile 2. So, this is uh certainly holding me over. If I am you, I try Vibe Coding now inside Hermes Agent. Especially if you have like your Chad GBT subscription plugged in. So, you’re so you get all those free credits with your subscription. I think you get get away with using this as your vibe coding tool now because you can see the diffs, you can create pull requests, all of that. Really, really nicely done. This desktop app is incredible. I mean, this is probably the sole reason Hermes is ahead of OpenClaw right now. So, take advantage of it. Test it out. Do some vibe coding in Hermes. I’m really, really impressed. And the last bonus recommendation I want to give you here is around Fable 5. Fable 5 is now available in Hermes Agent. As you can see here, they have Fable 5. Here is my recommendation on how to handle this. I would go in, start a new session, and ask your agent to create a new Hermes profile specifically for Fable 5. So, here’s the prompt I would use. Please make a new Hermes profile called Oracle that uses Fable 5 model from Claude. Let me know when that’s set up so I can use it in Telegram. For those who don’t know, profiles in Hermes are basically new Hermes agents. So, every profile you set up is a new Hermes agent you can use. I highly recommend setting up a new Hermes profile built specifically for the Fable 5 model. And I’m not using this as the daily driver. Not at all. You do not want to use as daily driver. It is way too expensive for that. But what I’d use it for is super super super complex problems. Right? You have your mixture of agents option which is one way to handle complex problems that I think is more for like complex questions and getting really wellthoughtout answers. What I like to use Fable 5 for is when I’m building things with Hermes agent. Hey, build this out for me, build out for me. I find coding with Fable is significantly better even when you use Mixture of agents. Mixture of agents is good when you need opinions from multiple models to get a nice synthesized answer. Fable is good when you want like a perfectly built out front end, a perfectly built whatever, Fable 5 will do that the best. So, I highly recommend or if you’re doing really complex actions across your computers, like if you want to set up a local model on your DJX Spark that connects to your Mac Studio that does this on another computer, really complex tasks that are multi-step across different devices, I would use Fable Agent for that as well. So your line of thinking is complex questions, use mixture of agents, complex tasks, use your fable agent. You set that up by setting up a new profile inside Hermes and it will go and build that out for you. Then you can set it up in Telegram and you’re good to go. You can go on your Telegram and talk to that agent at any time. Just as a side note, I’m going to do a full boot camp on Hermes agents. We have multiple full boot camps on Hermes agent inside the Vibe Coding Academy. Link for that is down below. It’s the number one community on AI on the entire internet. I promise you will not regret joining. Build alongside a thousand people doing cool stuff with AI. You’ll learn a ton of stuff. Promise it’ll be worth it. Link down below for that. You learned anything at all, leave a like down below. Let me know what you want in the next video. More Hermes agents tips, more Fable 5 tips, more Claude Code tips. Let me know down below. All my videos are based on your feedback. I hope this was helpful. Make sure to turn on notifications so you do not miss the next video. I will see you in the next

The world has changed dramatically this week and I am not understating that. It has never been more important for you to get into local AI. Governments, the US government is banning frontier models for everybody. Hardware prices are becoming outrageous. They are no longer affordable for the average person and they may never be affordable again for the average person. Chinese labs are creating models that are on unparalleled levels. the world is transforming right before our eyes. And I believe the most important thing you can be doing to keep up with all this change is educating yourself and getting into local AI. In this video, I’m going to cover why that is, how the world is changing so rapidly, what hardware you need to get into local AI, what models you should be paying attention to, what use cases there are for local AI, why would you use local AI when you can do Chad GPT for $20 a month. I’ll cover it all. you’ll walk away an expert and even maybe set up with a local model by the end of the video. This could change your life. I don’t think it’s ever been more important to pay attention to the video as it is this one. Just as context, I added this computer to my home AI lab this week. I built this 1590 RTX computer. I already have three Mac Studio 512 GB. I already have two Mac Minis. One of the Mac minis is not shown here. And I have a DJX Spark. Do you need to get all these hardware? Do you need to spend $40,000 like me? No, you do not. In fact, you might be able to do a really great home AI lab setup with the hardware you already have, with the laptop you got in your closet. I’ll cover all that in a second. But why is this so important? Why am I building up this entire home AI lab with all these different computers? Well, the world has changed a lot this week. First of all, Chad GBT56 was announced yesterday, and with it, it was also announced that only a hand selected group of people by the US government will be able to use it. That stinks. It’s a big party and you ain’t in it, right? You’re not able to use the two best models out there right now, which is Fable 5 and Chad GBT56. They’ll probably return within the next couple weeks will be able to use it. But the hand selected people by Anthropic, by Open AI, by the government will have a head start and they will have a massive advantage on you. That sucks. We have officially entered the age of hand selected winners, which isn’t good. On top of that, the prices of literally everything are going up. The prices of every piece of hardware is dramatically going up. Apple announced that the first of many price raises just happened. Many of their devices went up 20%, 25%. I mean, you’re not getting Mac minis for less than $1,500 right now. It’s absolutely insane what’s going on. I went to MicroEnter. I wanted to build this computer. I wanted to put 128 gigs of RAM in it, which isn’t the craziest thing to do in the entire world. It would have cost me $4,000 extra dollars. Dude, I’m already spending $9,000 on that computer. It would have cost me $4,000 extra if I want to do 128 gig. I mean, even the cost of like the PlayStation 5, the Xbox Series X and S are out of control. The Series S is $500 now, which was the cost of the good model when it just came out. I don’t know if you remember video game consoles, they used to drop in price. They used to be like $99 by the end of their cycle. Now they just go up and up and up. We are living in absolutely insane times. Hardware, while very, very expensive, is still obtainable. And that’s why I’m making this video right now is because hardware is still obtainable. I predict it won’t be obtainable. I predict in a year it’s going to be absolutely insane prices, way higher than it is now, and it’s already at insane levels. Two years, I don’t think it’ll be obtainable. We are still years away from humanoid robots being a thing. We are still years away from drones fighting in wars. Most people who are in the know are in agreement that every home will have a humanoid robot and humans will no longer fight in wars. It’ll be all drones. We are nowhere close to that. And every humanoid robot, every drone that fights in a war, every automatic self-driving car out there will need absurd amounts of memory. And those products aren’t even being built right now. And we have a bottleneck. So, what happens when we actually do start producing humanoid robots? What happens when we actually do start producing tens, if not hundreds of millions of drones to go and fight in wars? I’ll tell you this much, you ain’t getting memory when that happens. Memory ain’t going to be available to the normal average Joe when that happens. So, yes, prices are absurd right now, but no, I don’t even think we’re close to the end. In fact, I don’t think we’re ever going back to the days of before where these kinds of prices were obtainable for people. So we live in a situation frontier models being taken away. Hardware prices are absurd. What is the solution? To me, the solution is getting into local AI now. Getting the hardware for local AI now. Because when the government can start controlling who uses what models, you need to become sovereign. But the issue is is the window for becoming sovereign is becoming very small because the hardware is getting more and more expensive. So, I believe we’re in a small window right now where it is obtainable to get local AI. Even though it is very expensive, it’s still obtainable. That’s why I’m making this video right now. What is local AI for those who are really not in the know? And by the way, there’s chapters down below, I think. Feel free to click around. It is AI that runs on your computer. When you send prompts right now to Chad GBT or Claude or whatever you’re using, it goes to the cloud, right? It goes through your internet, goes to a server, goes to a data center. It’s processed in that data center. Then the answer is sent back to you. That is in full control of those companies. Your prompts live on their servers. The answers they give you are determined by the companies. You are fully out of control. When you run local, the models run on your GPUs on your devices. So on your computer. So your prompts, your answers never leave your computer. They don’t even leave your internet. You don’t even need internet. It’s all done locally on your computer. That’s what I mean by becoming sovereign. No one can take it away. It is completely private. So, let’s talk about those advantages. Again, as I said, private. No one can read your logs. When you send a prompt to Anthropic or Claude, if they decide uh to sue you or something like that, they can access your logs. They can read your logs. Anyone at those companies, the government, anyone can read everything you’re saying. So, those conversations with your AI girlfriend, they ain’t private. Anyone can read it. The government can read it. So, privacy, local AI, completely private, unlimited. You’re not paying a tollgate to send your prompts to the cloud. It’s unlimited. It is run on your computer. You’re just paying the cost electricity going into your computer and it can’t be taken away unless they come into your home guns ablazing and take your Mac Studio, your DGX Spark, take it away from you. They can’t take it away. It’s all yours at any given time. As we saw the last couple weeks, it can be taken away. They took away Fable 5. They stopped you from being able to use Chad GBT 5.6. You can’t use the best intelligence out there right now. They determine who can use it, and you ain’t one of them. Before we get into what hardware you need, what models you should be using, let’s just address the elephant in the room. The number one thing naysayers say to me, they go, “Alex, but isn’t local models stupider? Isn’t it dumb? Aren’t they slow?” Yeah, you’re you’re 100% correct. They are stupider than your Chad GBT models you get at Chad GBT.com. They are stupider than Opus48. They are slower, sometimes significantly slower. Right. But that’s not the point. You’re 100% right about that. But here’s the thing. One, not for long. I don’t think it’ll be stupider and slower for long. In fact, GLM 5.2, the newest local model, obviously from GLM, is just as smart as Opus 48. I’ve tested. It is just as smart as Opus 48. So, they are getting significantly smarter and they are getting faster. They are a bit slow right now, but these companies, including Google, are working to optimize these models so they fit on your computers. And that’s why the old paradigm of like hardware going extinct and not being relevant 2 to 3 years after you buy it doesn’t really matter anymore. That that kind of cyclicality of hardware doesn’t really exist anymore because these models are becoming more efficient. And the hardware from 5 years ago could start running Frontier models soon enough potentially. That’s why I think the excuse a lot of people make, oh, but won’t the hardware not matter in a few years? No, that’s I I I do think it’ll matter in a few years. In fact, I don’t think you’ll be able to buy more hardware in a few years. So what you got available today is all you can get. So little bit of expectation setting right now a little bit stubber right now a little bit slower. Even though it’s stupider and slower, there are use cases that only local AI models can do that Chad GBT and Opus can’t do that I’ll go in as well that are massive advantages that I really think you guys should be getting into. So let’s first talk about the hardware you should be getting. Can I use Mac minis? Can I use Mac Studio? Should I buy a Mac? Should I buy an Nvidia card? Should I buy a DJX Spark? Well, here’s the three categories of hardware for you. Actually, there might be a fourth one here. Let’s talk about the fourth one, too, which is kind of just dusty old computers/cheep computers, like kind of the budget computers. We’ll talk about in a second. Let’s first go into kind of the more expensive ones people get into. You have the Mac Studios. This is the kind of one I started a big trend on back in January when I first started talking about this. Mac Studios have what’s called high unified memory and low bandwidth. What does that mean? Macs, any Mac computer you get, the advantage of the Apple architecture is it’s unified memory. So the RAM and the VRAMm are kind of one thing. The RAM and where you load models into models need memory are combined, which means you can load much bigger models. You can load I load GLM 5.2 on one of my Mac Studios. It’s a 250 gigabyte model. You need 250 GB of memory. That’s a tremendous amount. So, you can load massive models, but it has low bandwidth. What does that mean? Low bandwidth. That means it can’t process a ton of that memory at once. So, they’re really, really slow. So, you buy a Mac, you can run very smart intelligence, just not going to do it too fast. Then you got these AI computers, these kind of AI workstations that are now coming out like the DGX Spark, the AMD Halos, the DGX station. They have kind of medium unified memory. DJX Spark has 128 gigs, which is really good. Max Studio. I have 512 on these, which is obviously a lot better, but they also have medium bandwidth, so they can do it pretty, you know, a good amount faster than the Max out there. So, that’s pretty good. You can run pretty smart models at pretty good speeds. And then you have kind of the powerhouse chips, the really expensive ones, the 5090s, which is what I just built, and the 6000 Pros, which is now a $12,000 GPU. They have lower VRAM. The 6000 has 96 gigs, so actually has pretty good VRAM. But the 5090 uh has 32 gigs of VRAM. So lower VRAM, but very, very high bandwidth, meaning it’s lightning quick, very, very fast. You can run pretty good actually, you know, for 32 gigs, you can run good models now, and you can run them very, very fast. Those are kind of your three options. Then you have the fourth category, which is basically everything else. Your Mac minis, your old laptops, you know, the pretty cheap computers out there. or anything below $4,000 computers, you can run some smaller models out there, some Gemma 4s, and they are solid models. You’re just not going to give them any sort of powerhouse use cases. They’re not going to write any sort of kind of quality code for you, but they can do some small processing. They can do embeddings, which is improving basically the memory of your agents. They can do some good things. So, even if you just have a Mac Mini, you can still take advantage of local models, and you should be taking advantage of them. So, it’s up to you what kind of functionality you want. I’m going to go into use cases in a second, and that’ll might help you decide what hardware you want, but this is where you’re at. Also, for just from a usability perspective, I personally like Mac OS better than all the other operating systems. I love it. I got an iPhone. I got an iPad. They all integrate together. So, that’s why the Mac Studio is kind of my main driver. AI computers like the Spark, they’re running on Linux, right? The usability for the average person is not great. So, you can still get them. They’re just not going to be very usable. you’re probably going to want your Hermes agent to run everything. And you have the powerhouse chips. You can run on Microsoft Windows. People like Windows. People are used to Windows. I personally kind of despise Windows, but you have that option as well. I’m going to get into use cases in a second. I’m going to show you some things I’m doing. But last thing I’ll show you is like kind of what else you need, what software supports the hardware. I’d say two things are critical here. I’d say tail scale. Tail scale is a piece of software that allows you to create your own private network amongst all your devices, right? So, I have like seven or eight different computers here. I have tail scale on all of them. So, they’re all on kind of the same closed network. And no matter where I am in the world, I’m on my Wi-Fi, not on my Wi-Fi, all my computers and the models can talk to each other. This allows you to leverage the local models you’re running on the devices on any device you have. You if you run a really nice local model, your Mac Studio, if you’re on the tail scale on your phone, you can run it from your phone anywhere in the world. Use that model locally on your phone, which is really, really nice. And then Hermes or OpenClaw. This is the killer one. This is one you need. This basically takes away all the need to have technical knowledge. I just have my Hermes on my main Mac studio. And because all my devices are on tail scale, my Hermes can control, load, do whatever it needs with the local models. I can go to Hermes. I can say, “Hey, go on my Mac Studio, load Quen 36 onto it.” It’ll go over to my Mac Studio through Tailscale, load the model, make sure it’s working, do all the work for me. It’s basically like my my IT guy. I say what I need. I say, “Hey, I want to write this piece of software. Use the model that’s on this computer. Make a route through this computer and use this other computer and this other local model do this.” And it’ll just figure it all out and put all the pieces together. Hermes and Open Claw are absolutely critical for the setup. You need both of these. And the good news is they’re both free. Now, let’s talk about use cases. This is very important. And everyone goes, “Man, why would I use a local model if I can just use chatgbt.com for $20 a month and it’s way smarter?” Well, here’s the thing. You open up a whole new paradigm of use cases when you have always on ambient 24/7 AI because the models are unlimited. They’re always on. There’s no rate limits on your computer. you unlock a whole new class of use cases because they can run 24/7 and work around the clock. I’ll show you a few things. This is my home AI platform I built out. By the way, if you learned anything so far, leave a like down below, subscribe, turn notifications. Also, I have a community, Vibe Code Academy. Feel free to join that. I do live boot camps like three times a week in there. So, if you want to join, ask me questions. I’ll walk you through setting this up. I’ll handhold you through it. Join down below for the Vibe Code Academy link down below. But this is my home AI setup. I’m building this platform out myself. My Hermes agent built this out for me. But I want to show you some things here. This is saying failed, but it’s actually working. I am doing multiple things around the clock 247. So, first of all, you can see my running models. On my Mac Studio 2, I have GLM 5.2, which is Opus Level Intelligence. It’s a little slow. On my DGX Spark, I have Orith 1.0 setup. I still need to set up my RTX 5090 computer. Built that just yesterday, but I’ll be running probably another copy of Ornith 1.0 on there. Very cool local model that just came out. And they’re doing many things all at once. One, they’re doing security scans on my codebase. So I’m building a SAS right now called Henry intelligent machines. It is doing around the clock security scans on my codebase, looking at different APIs, different pieces of code, making sure it’s secure, making sure nothing’s going on. It’s taking a look at my database. It’s looking for anomalies in my database. This is going around the clock 24/7. You can’t do this with cloud models unless you’re filthy rich because you’re paying a toll booth for every API call to the the cloud models because they’re local and they’re unlimited. I can have it ambient. I can have it working at all times around the clock all day. It would cost me one of these computers to do this with Claude Opus 48, but because it’s unlimited local, it works out for me. I have another thing going on right now too, which is one of my local models is at all times of the day scraping Reddit, looking at tweets, pulling down posts from X, and it’s looking for business opportunities, opportunities to build SAS, opportunities to build services, software, whatever I need. It’s constantly scraping and scanning the internet to find these opportunities. Now, these opportunities are going into my SAS Henry intelligent machines. My SAS is kind of built around fighting opportunities, but I also take a look every once in a while and I look to see what kind of software I can build, right? Like look at all these bit every day, every couple hours, I get a list of opportunities with sources from people’s tweets saying, “Hey, this person asked for this question or has this problem and it shows me the pain. It shows me how to solve it and it ranks the opportunities for me.” And this is happening again around the clock every 20 minutes just watching the web, looking for signal, looking for opportunities. Again, not possible with cloud models because it’d be way too expensive to do this to run it around the clock. But because I’m running local models, which are unlimited, I can do these types of things. So, I basically have a fleet of employees working around the clock 24/7. Never need to eat, sleep, do nothing, never ask for time off, doesn’t need health insurance, and it’s only possible with local models. Those are the unlocked use cases. And this is just now, right? Think of what happens when these models get more efficient. When Google does better research on how to run miles more efficiently locally, you’ll be able to run GLM 5.2 at rapid fast speeds at cloud speed. So you can have frontier intelligence at amazing speeds and have unlimited of it. This is how you escape the control of what’s going on right now. You escape the control of the frontier labs. You escape the control of the government who’s holding everyone back. This is how you escape the I I hate to use the cringe term. It’s like the most cringe guru term on planet, but this is how you escape the matrix. This is kind of it. I also say this, and this is the most underrated part people don’t talk about. Two things. One, it’s just fun. It’s fun building your own hardware. It’s fun running local models. It’s fun. You’re allowed to have fun. Believe it or not, in this world, you are allowed to have fun. You’re allowed to experiment and tinker. It’s never been more important to experiment and tinker. So, one, you’re allowed to have fun. Two, it’s educational. You’re learning about the most cuttingedge important technology in the history of our species. Literally that it literally is. And the best way to learn about it is by getting hands-on and doing it and running yourself and seeing what the capabilities are. Right? So, it’s one helpful cuz I showed you all this use case that made my life significantly better. And two, it’s fun. You have a lot of fun doing it. It’s fun learning about the hardware. It’s fun running it. It’s fun booting it up. If you’re anything like me, you can, you know, I built this 5090. I’m running local models. Hey, I’m also playing Cyberpunk 2077 in ultra mode. So, that’s pretty fun, too, just as an aside. And then three, educational. You’re learning. Education is important. You’re learning about the most important technology ever. Now, if you set it up like I set it up where your Hermes is controlling everything, they’re tail scaled out the wazoo. You literally can build anything you want. If you just want to build a simple chat interface that looks exactly like chat GBT, you can go to your Hermes and say, “Hey, build me a chat GBT interface.” If you want to give your local model a face and a name so you can talk to it and you have your own buddy and your own pal living on your computer because hey it’s kind of fun looking down your computer knowing there’s an AI agent on it. You can do go Hermes. Hey build a face my local agent put on top it whenever it sends me a message back make it talk. You can do it. You can do whatever you want. It’s complete freedom with the most important technology ever made. I think it has never been more important with hardware becoming about to become unobtainable and the government taking away uh frontier models. I don’t think there’s ever been a more important time to get into this technology. I hope this was helpful. Leave a like, subscribe down below if you want. Let me know in the comments. You want more local content. You want me to talk about the home AI lab setup? Where where are you most confused? Where can I help out the most? All my videos are based on your feedback. Let me know down below. Hope this was helpful. See you around.