Running a single AI agent is great, but running an entire team of them is a completely different level. So, in this video, I’m going to show you how to run a company of agents. We’ll create a team together, give each agent a role, put a manager in charge, and let them run real work on their own. And the best part is I found a platform where we can do this with absolutely no code required. Let me show you how it works. Let’s start with why the one agent thing breaks. I mean, you’ve probably felt it if you’ve ever tried giving one agent hundreds of different jobs. You ask it to do research, writing, and design, and what ends up coming back is, let’s just say, mid. It’s the same reason you wouldn’t hire one person to be your researcher, your writer, and your designer. They’d be stretched too thin to be good at any of it. But, honestly, that’s the small wall. And there are different ways to solve for this, which I cover in some of my other videos. However, there is a bigger wall, and nobody really talks about it. Here’s the thing, I don’t just play with agents. I actually run a business with a team of nine people, and every AI agent harness I’ve used, whether it’s Hermes or Openclaw, they’re great single-player agents. I mean, they are built and designed to be used for one person. That’s not a bug, it’s a feature of those tools. So, the second you try to run that like an actual team, it starts falling apart. Your people can’t get in. They can’t see what the agent did. They can’t jump into the same thread. They can’t work alongside it. And most companies are not single-player. I mean, we’ve tried adding a Hermes agent to a Slack team channel, but permissions on there are a nightmare. And that was a big wall for me. I have this powerful agent and a team of nine people, but no way to put them all together in the same room. So, let’s fix it the way a real company would. And that’s exactly how I found HyperAgent. It’s the first setup I’ve used that’s actually built for teams instead of one person. So, my whole agency can work inside of it together. And luckily, this video is sponsored by HyperAgent. HyperAgent is a place to hire and run a whole team of AI agents in the cloud with no code required. Nothing to install. You just sign in, make an account, and you’re in. Oh, and you can now sign in to HyperAgent with the ChatGPT plan you already pay for. So, you don’t have to pay for any other AI subscription. And if you don’t have a ChatGPT plan, you can use my link below to get $100 in free credits to use on any AI model that you want. Okay, let’s hire our first agent. Building an agent in here is basically like writing a job posting. To hire our first agent, we’re going to click over here, new agent, and the first step is identity. We have to describe what the agent’s job is. And I’m going to start with our manager agent. He’s going to be in charge of managing and orchestrating all the other agents. So, let’s write, you are Matt, chief of staff, you run my company’s AI agent team, and you report only to me. Let me actually add, your job is to take what I ask for, break it into the right pieces, and hand each piece to the specialist who owns it. You don’t do the work yourself, you delegate it, you follow up on it, and you bring me back one clear answer. Always keep it short and concise. I hate long paragraphs of fluff. That’s it. From there, we’ll click on draft identity, and it will create its own system prompt. We, there you go. It just did. It gave itself a system prompt, a description, and we’re going to give it a name. I’m going to go for Max. And then, the next step is picking out its model. Since I connected my GPT subscription, I’ll go with my OpenAI, uh, not GPT-6 Astra. We don’t need something that strong. I’ll go with the latest. So, bigger models make better calls and cost more. So, match the model to the role. You can even set a budget cap for each and every one of them by toggling this on and then putting what your budget limit is. Since I’m on the subscription, I’m going to keep it as is. And then, when we’re setting up the other agents, you can point the grunt work at a cheaper model. Then, we go to tools and integrations. These are essentially what these you want to hand each agent. For example, our content agent that we’ll create later needs to have Instagram access. So, we’ll connect that. But, I don’t think our chief of staff necessarily needs that, so I’ll just go to the next one here, which is invocations. Here, you can choose where other places you want to be able to reach your agent. If you want to talk to it via Telegram or for it to talk to you via Slack or email, you can turn that on and set that up right here. I don’t need that necessarily. I want to be able to talk to it here, so I’ll just keep it as is. Oh, and another thing you can set here before I forget is who can also reach your agent. I which I’ve set this to anyone since we’re going to include my team when I set this up as well. Then, finally, the last step is knowledge. If you’d like to customize this, you can, but I’ll keep this as it is by default and click on create agent. And there you go. We can also give it a an image, which I would like to do that. I think Max is a wizard. I’d like a 16-bit avatar of him. And it’s generating it. And there we go. I love it. I’m going to choose to keep it, save it, and from there, the next step is to create a new thread with our agent and make sure that plan mode is on. It is on by default, but just make sure. That way, it’ll show you its steps before it touches anything for the first turn, for the first prompt we’re going to give it. And with that, let’s hire the rest of its team members. Now, we bring on the rest of the company. There are two ways we can do this, either by going through what we just did for every different AI team member, or by asking our chief of staff agent to do it for us. And that’s exactly what we’re going to do here. So, I’m going to send it. Before you can run anything, you need a team, so let’s hire them. Create these four agents. Number one, a prospector, who finds and vets new leads every day and hands the qualified ones to the setter. He needs web search and browser. Then, second, we need a setter, who lives in my inbox, replies to incoming leads, and drafts outreach in my voice. I already have a template to reply. We can set for it. Never send anything without my approval. Needs email access. Another tool we need to make sure this specific agent has. Then, is a client success manager, who keeps an eye on our active clients and their feedback and flags anyone who’s unhappy or gone quiet. He needs access to where our client conversations live. I already have a bot set up for that, or rather, a Telegram registry set up for that, so we need to link it to this. And then, a content creator, who tracks my Instagram performance and brings me content ideas and full drafts to film. Needs access to my Instagram, its analytics, and to VidIQ. The purpose of you and your team is to help me manage and scale my business. This is the first prompt I’m going with. I’m going to hit send. And just like that, it’ll open its first thread in plan mode for that first prompt only, where it’s going to start ideating and thinking and going over the first prompt that I sent it, what I like to call the startup prompt. And from there, we’re going to see what it’s going to come back with. So, here, it already asked the first question, which inbox should the setter work from? I use Gmail, so Gmail. Where do your active client conversations and feedback primarily live? Multiple places. Let me actually name them. Telegram plus Slack. Telegram plus Slack. I already have an API that I can give you later on to access this data. And what should I use to configure the setter’s voice and the content creator’s research workflow? I’m going to send that. Those are the three questions it asked me with their three replies. Sent me the team structure is clear. I’m turning it into an implementation plan that pre-preserves approval gates and accounts for the integration that still needs to be connected. So, starting to draft the implementation plan, and it asked me one more question, “How would you like me to proceed with this four-agent team plan?” Let’s read it. Yep, it seems to have understood well what I want, so I’m going to approve its plan and send that, saving all the things that I’ve given it to memory, and it’s going to continue to do that every time it learns something new. It’s very good, and we’re able to see what it’s saving. There we go. These are the four agents. All I need to do, I can look into each one if I want to. See prospector, there it is. Integrations, even set its schedule for it. I save all of those agents. I’m going to tell it that I saved. Here is the next step, which is connecting each agent to whatever tools they need. Here, really, it’s very customizable depending on what your own specific agents that you’re creating need. What you can do, and what we can see here, is all of them appear on the left bar. So, what I’m going to do is I’m going to click on each one and make sure that they have their capability set. For example, the content creator needs access to Instagram and to VidIQ, so I’m going to go ahead and connect everything that I need to connect. And there we go. Just a few minutes later, everything is set up. I’ve even given each of them a profile picture and all the capabilities. Like, the setter has access to Gmail, the content creator has access to Instagram. And by the way, there’s a whole ton of integrations. I doubt that you won’t be able to find whatever integration is that you want. So, now we’ve got a company. Let’s put them in the same room and start working. It’s called rooms, and you can access all of them over here. A room is basically the office. It’s like a Slack channel where you can add different agents and different team members to different groups. It’s essentially one shared workspace where your whole company works together. And three things happen here. Your specialist agents, in our case the prospector, the setter, the client success manager, and the content creator all run in parallel instead of one at a time. And then your actual human teammates can jump into that same thread and talk to them. Let’s call this one general chat and create the room and most importantly add our members like setter, the content creator, the client success manager, the prospector, and most importantly our operator Max. And I’m going to send hi Max, I’ve just set up this room. This is where we’ll all be working together. And Max answered hi Charbel, got it. I’ll use this room to coordinate with you and the team, keep updates, and pull in the right specialist when needed. Max, what else do you need so that we go live? I’m ready. And here you can see them start to show up. I sent the message to Max, but because the setter is involved, I’ll need the setter to draft an outbound template. That’s literally what I said. The setter is even typing. Like imagine this, the prospector drops a batch of leads into the room, the setter picks them up and drafts the replies, and then it tags me, the human, with something like hey, want me to send these? And nothing goes out the door until a person says so. Or depending on your permissions, I can set it to run automatically, but for the first drafts, I recommend that you keep human approval. And that right here is a team stand up running on its own. So, let’s give them a real job and watch what happens. This is amazing because I asked Max, our operator, to go ahead and start the workflow, and it came up with a plan, and then it started delegating. Here you can see there 12 replies. I sent one message, I received 12 replies, and that’s not them necessarily talking to me, that’s them communicating with one another. And we can see that. The setter sent that it found 15 inbound leads along with an email draft that it gave me, which I do want to change, but it’s good because this is where we get to set it up, and an outbound email template, which I absolutely love. This one I’m not going to change. I’m actually going to tell it. And it also sent us the client success manager report as well, which I’m going to blur for obvious reasons, sensitive client information. But I’ve told it that’s great. Let’s set it up so that the client success manager sends us a ping around anything that’s urgent every single day. Look at this. It’s amazing. Max is tagging the client success manager. Hey, don’t send anything. The client success manager, yeah, of course, everything is read-only. No client messages are being sent. Look at that. Even the content creator came in the loop and sent us his report essentially. And look at that. The prospector came back with eight founders found. And I’m going to ask Max every time the prospector finds a lead, they should have an email. If they don’t have an email, then we can’t send anything. And if they have an email, then that to the setter, which will send our outbound template. So, to recap, by just telling them to go, after we’ve outlined everything that we’ve outlined, the prospector found and vetted new leads and handed them to the setter who drafted replies. Then the content creator came in, read my Instagram numbers, and pulled together today’s content ideas with drafts ready in my voice. Look at this. Before you ask AI for 30 content ideas, open your last three sales calls along with the full script, what visuals should be on there, and a CTA and captions as well. We have one, two, three short-form video ideas. And we have our client success manager that scanned clients for us and flagged one that’s gone quiet. Then Max took all of this and packaged it for me very tightly and nicely so that I don’t need to go through everything I just went through. Tagged me only when it needed something from me for me to approve. Absolutely love this. What’s very impressive is I didn’t tell it how to find outbound leads. But looking into its data, it’s literally pulling in different sources for every single lead. No two sources are the same. And it’s going to find their LinkedIn. From their LinkedIn, it’s finding their emails, and it’s drafting the emails with the information it finds. Do you start to see the vision of how a company humans and AI starts to work and click together? And you think about this because the prospector goes and hunts for new leads, outbound leads. And he sends that to the setter, who then approaches those leads. Just like this. I click approve, and there we go. We’ve just sent our first outbound lead that we didn’t even have 5 minutes ago. That lets us set calls and close them. And then our content creator goes out and comes up with content ideas for us for Instagram. Which by the way, if you’re not following my Instagram, you absolutely should. I just started posting content on there, and I absolutely love it. There’s a ton of great free value on there. It’s Charbel XYZ. But imagine that, the content creator drafts you ideas for your Instagram. You post that content, and you get inbound leads messaging you on your email. The setter replies to them, also helps you close the deal, and then the client success manager helps you keep track of all your clients, making sure that all of them are happy and satisfied. That’s the entire vision of how HyperAgent can help inside of that organization. There we go. It just sent the first email. This is magical. All right, now that you’ve seen how this all works, you might be thinking, “I already run an agent. I’ve got Hermes or Open Claw or any other AI agent harness already. How is HyperAgent any different? Or how does it compare?” So, let me break it down. It’s not about who can run more agents because most of them can run a few. The real difference is who they’re built for. Almost every one of these is single player, which is great if you work alone. But I don’t. I’ve got a team of nine, and HyperAgent is the only one I found that’s actually multi-player, where my whole team works in the same space, in the same room, right alongside all of the agents. That’s the gap nobody else is filling. Hermes and Open Claw I genuinely love. I’m a big Hermes agent super fan myself. If you’ve noticed or taken a look at any of my channel, those tools are open-source. They run on your own machine. You own them completely. But, they’re built around a single always-on agent doing your jobs one after another. Grok Bot is multi-agent, starts at $100 a month and it’s hard to get a team inside. OpenAI’s dots is the same idea. It lives inside ChatGPT, but it’s also gated to only pro users, so also $100 a month. Meta’s Muse is a free single assistant for the everyday stuff, booking, shopping, email. HyperAgent is playing a different game. It’s the one built from the ground up for a team. Many specialists, uh manager over them, running in parallel in a shared room with you inside of it. It’s not a better single agent. That’s an org for your team and agents to work inside of it. So, here’s where I land. If you’re solo, a single agent takes you surprisingly far. So, start there. But, the moment you’ve got a team and clients and real work flowing, one agent in your own private window stops being enough. You need your whole company in the same room, your people and your agents together. That’s what I was missing for months, and it’s what made me add this tool into my toolbox. This was never about running one more agent. It’s about running your business like a company that keeps moving without you. From doing everything yourself to running the whole thing. So, tell me in the comments, what’s the first use case you’d use HyperAgent for? I want to see what everyone builds. And if you want to build your own team, HyperAgent is giving you $100 in free bonus credits when you sign up through the first link in the description. So, go hire your first agent and build it alongside me. If you enjoyed this video, then leave a like. Or, if you’re new to my channel, then subscribe because I’ve got a ton more content like this coming your way. Oh, and would you look at that? The algorithm gods really seem to think that you’re going to enjoy this next video as well. So, click it and I’ll see you there.