How Claude AI Runs My Whole Agent OS

**Julian Goldie** (0:00)
Today, we're going to be answering some of the latest questions about the agent operating system, how to build them, how to use them, how to get the most out of them, and some of the cool things you can do with this stuff. Now, I want to get straight into this, but basically these are questions from our AI Profit Board and community, and I answer them every single day in a video tutorial like this. Number one day with the members, and number two, because I know that if they have those questions, you probably have those questions too. And so if you want to ask me any questions like this, and I'll create a video tutorial like I'm doing right now for you, then you can post questions inside the community, link in the comment description, or go to the aiproperborn.com. Let's get straight into this. We got a few questions about Hermes, so we'll kick it off over there. So shout out to Maren. I have several questions that hopefully the community is running to and can recommend. Number one, how to set up and create the migration processes? Is Docker or Linux instances the best way to start? So number one for this question, I'm not 100% sure what you mean, but basically, if you want to set up Hermes Agent, you can set it up on a VPS if you want to keep it isolated, or you could set up with Docker as well. For me personally, I run it locally and I don't have any issues. You can also run it with, for example, like, Tailscale, so you can have, for example, the agent operating system with Hermes Agent plugged in, and then you can actually have that connected to Tailscale, so you can control it from mobile as well. And that way it's on a secure environment via Tailscale, but you can still access it on multiple different instances, and you can migrate that instance to multiple different devices. The other option that you have for connecting with Hermes across multiple different places is you can just connect it to Telegram or Discord and control it from your phone. Question number two. Is there a course or template to use if I want Hermes as the top orchestrator and Chrome manager, and then sub-agent spawners and Claude as technical architect and developer fixer roles?
Yeah, so what you can actually use is Paperclip. The Paperclip basically orchestrates your agents inside a team. So if you want separate agents, maybe one is the CEO, one is the architecture developer, one as, for example, the sub-agent that does all the grunt work, then you can use Paperclip. Now, Paperclip can come in two different formats. So we have it inside our agent operating system, but then you can also use just the standalone GitHub projects for that as well. The other thing that I'd say is that Hermes now has a delegate tool, which means it automatically spawns sub-agents for you. So you don't even need to manage them yourselves. If you're just using Hermes, it will automatically do it for you when you ask it. Last question, which is also, would having a Hermes master agent with separate profiles in each environment or separate instances in local machine and master orchestrators outside the spawn sub-agents in each environment work best? Complicated question.
So, for me personally, I will keep it as simple as I possibly can. Now, if you wanted multiple different Hermes agents working together on separate instances, all in the same environment, you could just have like a Telegram group chat or you could have a Discord group chat with all the Hermes agents added in, or you could, for example, even use something like Buzz. So, Buzz is a new free open source app from Jack Dorsey, and that basically allows you to add other humans and other instances of other agents. So, you could have them all working inside one environment. But again, for me personally, I had to keep it simple. I just wouldn't do that because I think it's complicating something that doesn't need to be complicated. Now, we actually got a question about the setup of our agent OS. So, for example, let's say you're using Hermes, and let's say inside the studio, we've got Minimax and Grok set up, but you don't have Minimax or Grok and you don't want to use them.
So, when you set up the agent operating system using the ZIP file, you can ask the agent that sets up the agent OS for you to use different models or images or anything else that you're creating inside there. And literally all you do is you ask the agent to swap it over and then test it and make sure it actually works. So this way, the agent, for example, like Claude or Hermes or whatever agent you use to set it up, does all the hard work and the testing for you and the technical stuff.