NEW Agent OS just got 10x better...

Imagine having an entire team of AI agents, a builder, a designer, a video maker, an SEO expert, all working together all day, all night inside one powerful system. That's the agent operating system and today I am answering the top questions from our community so you can get even more out of it. I'll reveal the one model swap that makes everything run faster, how to run this whole system from your phone or even an old laptop and the free AI coder inside that builds for you, private, no wi-fi even needed. And stick around because later I'll show you the loop mode so you can set one goal, walk away and your agents keep working until it's done.

This is going to save you serious time and put you light years ahead. Let's get into it. Today we're going to be answering the latest questions on the agent operating system and this is basically a powerful system where you can have all of your agents working together. You can basically build and automate anything and it's a really powerful system that just improves every time you use it.

It's also got a memory setup inside there. We have custom workflows for example like open montage and open design and also you can swap out the brains and plug in different models as they get released. There's all sorts of cool stuff inside here as well like even a video agent, a music agent, an SEO agent and a loop engineering system as well. So anything that you want to automate you can do with this system.

We even have local and free AI coders inside there. So with that let's get straight into some of the questions from the AI Profit Board and this is where I answer the latest questions from the community. I'm going to get straight into answering them and helping you as much as I can because I know if everyone else has these questions you probably have these questions too. Let's get straight into it.

So one of the first questions we got here was from Jerome which is has chatGPT made rate limits on Hermes use? So for me I don't think there's any specific limits on Hermes use but I do think that Hermes will use up a decent amount of tokens depending on what tools you've got linked in there. So you can also switch the model if you want to avoid getting rate limited as well. So for example if you're using like GPT 5.6 sole with Hermes agent that's going to use up and get rate limited a lot quicker than something like Terra.

And also you can also switch to a blank slate profile with Hermes agent. What that allows you to do is basically reduce the amount of tools that you have inside the profile so that it's faster to respond and also uses less tokens. I also think the one of the reasons that I stopped using Codex in chatGPT versus Claude was it just gets rate limited way faster. So it tends to run out of tokens way quicker than anything else I've seen.

Another question we got here was could we use the agentic operating system with an M4 Mac mini? Yes I actually have a Mac mini as well that I've used the agent operating system with and it works really well. So you can use it I mean it's pretty lightweight like even a basic laptop you can use with this system here. And the great thing about that is it's just you know it's running off for example Hermes it's running off APIs so even for example like an old laptop you could run this system with.

There's actually people I know who are running this on their phone so they have for example Tailscale set up with their agentic operating system and then they can control the whole system from a mobile device or even from an iPad or something like that. So you don't need like a fancy setup. The only thing that you might need a better setup for is local models. But honestly local models are not that great and if you do want to use local models the one that I'd recommend that I've tested recently is LFM 2.5, 2.6 B.

That is the fastest best model I've seen work on a Mac Studio or anything Mac related. It just dropped yesterday so would recommend checking that out. And then you can use the local engine over here for coding and building stuff out and the great thing with the local engine is it's free, it's private, it can run offline, it can run without Wi-Fi and additionally anything that you build you can preview and everything is saved inside your workspace over here. So there's a lot of good features you can use with local models and also you can use LFM with Hermes agent because it's so fast and it works pretty nicely as well.

Next question we got was from John Meza here. He says what's your experience with Opus 5? So he says he switched to Opus 5 after Anthropic significantly reduced its price to compete with OpenAI's 5.6 model but his experience has been less than satisfactory. I know what you mean, I'll come to that in a second.

So although Anthropic describes Opus 5 as delivering a near-fable experience I have found it disappointing. He says you know frequently makes errors, inconsistent experience, pretty bad right. And you can see actually here if we look at the token usage it uses a lot more tokens compared to other models apparently according to this chart. So I've definitely seen the same thing and I switched back to Fable 5.

I actually felt like Opus 5 was good the first couple of days it came out but one thing I've seen since then is that it gets distracted easily, it seems to be super slow, it takes ages to do things and so one of the biggest issues is it uses up a lot of tokens whilst at the same time taking ages to do the tasks that Fable might be able to do you know three times faster for example. So I definitely agree with you, Opus 5 not a great model, super slow, great on benchmarks, great actually building, it just takes way longer, it gets distracted and uses up too many tokens. You can use open source projects like for example RTK or Caveman to reduce the amount of tokens you use but considering this is Opus 5 level model you shouldn't have to worry about that. However those are a couple of tactics you can use to reduce the amount of tokens.

So a question here from Ney which is do you use loops? So we actually built out a loop engineering system inside the AgentOS and we've actually got a couple of options. So number one is a loop engineering system under the loop section and you can define your definition of done, you can define a starting point if you already have something ready built, then from there you can select which model you want to use so you could even use a free model for building out and then you can select how many rounds it loops round for autonomously and also which model you want as the judge to define whether the work is done or not. And then you also have a workspace that you can monitor whether it actually looks good or not and whether it's been built.

So pretty cool system and then the other loop system that we have is the goal mode. So for example in Hermes and in Codex we have goal mode and basically you can set a goal and then it will just loop around for like 20 rounds or until the work is done. So those are two different methods you can use to reduce the amount of prompting you have to do and just set Hermes on a loop directly. By the way if anyone wants to grab our agentic operating system you can grab it inside the classroom, go to the new daily update section and you can find our video tutorial, the last update date and the zip file to install it.

Also getting some great reviews you know it's a really positive community you can see for example Rye is posting about how good the lessons are and the teaching style and everything else so thanks so much and really appreciate the feedback. And that's basically it for all of the questions inside the community today. Every single day I answer questions like this so if you have a question for me, you want me to make a video tutorial for you, maybe you want me to walk you through something or answer a question that you're struggling with or even set up a custom automation for you, you can post inside the community and I'll be happy to help you. And then we also have people online 24 7 there's 3,800 members in here so you can get help and support whenever you need it.

Inside the classroom you can get access to all of my best trainings on this sort of stuff and we've got a complete beginner to expert section as well as new daily updates inside this section right here. Inside the calendar you can also jump a weekly coaching course, ask questions, get help and support in real time, share your screen and then inside the map you can meet people in your local area who are building out with AI agents like you. And you might be thinking oh this is technical to use but actually you can see that I'm non-technical and I can use this and also there's over 207 pages of testimonials and wins from community members inside the AI platform so I know if all of us can do this then you can do it too. You don't need to be technical and you don't need a lot of time to to really master AI.

So obviously inside there, link in the comments description or just go to the aiprofitball.com. Thanks for watching.