NEW Agent OS is INSANE! (Runs 3 Businesses FREE!)

**Julian Goldie** (0:00)
So we're getting loads of questions about the agent operating system that we've built inside the AI Profit Boardroom community. I thought I'd answer them today because probably if they have questions, you have questions too. And so we're gonna run through some of the top ones today and talk through exactly how we use our agent operating system to basically automate anything and also to have multiple agents working together, all from one single shared memory and one single shared system that works together. And this is the first question we've got here from Conscience. So Conscience is saying, you know, how are you running multiple agents inside the agency? So is it designed to support multiple different agents at once?
How are you using multiple agents in practice? Have you successfully run several agents at once? Are there settings and modifications that you can do? And how do you personally define what a multiple agent is? So a lot of questions in one there. Let's run through them. So usually if I've got separate agents, it depends what we're doing. But if I want to run multiple sub-agents working together as a team, then the way that I do that is I have the Kanban board with Hermes, Agent, and then I can build multiple tasks together. And I can just plug in one prompt and then it goes off and does it, right? So we can plug in one prompt here. It will assign it to my different agent profiles with Hermes. So we've got like GLM Judge, GLM SEO, GLM Researcher, GLM Writer, et cetera. And then they can go off and just build whatever we want. And so that's pretty powerful because it can like write multiple articles or create videos together as a team in one single place. Now, each agent tag inside the agent OS has its own independent process, right? So for example, you've got Claude and then you've got Hermes, you've got Groq, Kimi, all running in their own backend session. So they genuinely run side by side. Switching tabs doesn't stop the others. The work like keeps going in the background. The UI just shows the tab that you're looking at.
So if an agent stops when you switch, that's usually an older build or something like that. But yeah, you can really have multiple parallel agents working. And again, Hermes have just shipped a new feature called asynchronous sub-agents. And so basically what this means is they have a new delegate tool that fans out workers in the background and no longer blocks what you're doing inside the chat as well. And so you can have one lead agent fanning out multiple different jobs to multiple agents at a time as well. So that's another way to use it. But this is the way that I look at it. Next question, AI video generation, right? So RK is asking about this and basically like how to generate videos with AI in a way that doesn't take up too much resources and how to organize it, how to get it all set up together, et cetera.
And what's like the best route for this? So here's the way that I look at it. If you want something that's kind of like less resource intensive, then you can use something like Groq or Wolf or Minimax. So for example, we have this studio here. We can generate videos and images and voice notes using Groq or Wolf or Minimax or Wolf. And we can switch between them inside the top here. So that's a good option. And then you can also generate Hyperframes as well. And I know you've put there the Hyperframe seems expensive, but it's a free open source skill. So you can give your agents access to Hyperframes and then it can go off and create videos for you. So if we have a look here, for example, this is a section we've created inside the Agent Operating System just for videos. And you can see that it can create like videos, you know, pretty nice animations, like as you can see here. And it can also, for example, create AI avatar videos if you plug that into HeyGen as well. But if you just want like, you know, videos without an avatar, then you can actually do it for free with your AI agents. So you got two options there, Hyperframes or you can use the studio with Minimax or Groq and Hermes. And if you're already subscribed to Twitter, actually, then you're not paying anything extra because you get Groq access already and you get Groq build access too. So this is pretty interesting. John was posting about basically finding a replacement to Perplexity Computer, which is Hermes agent.