how to run Agent OS for FREE (daily update!)

So it's time for another daily update. And if you're not sure about this, I'll link to it inside the description. But basically what I do here is just update you on what I'm working on and also just answer the latest questions from my channels in terms of like the questions we get and that sort of thing. So we'll just hop straight into it. And, you know, if you've got any questions, feel free to leave your comments on this video and I'll help you as much as I can. So we've got one question here, which is about agent operating systems. We created this video about agent operating systems with Claude and Alex says, what's the cost of this? My worry would be expensive usage, which is a great question. So we actually have a bunch of strategies to get the cost down if you're worried about that. So what you can actually do with an agent operating system, and I'll show you some examples, is you can actually use free models. So for example, we've got OmniRoot over here. We're also using HY3 Coder inside there. And also you can use local models. So the good thing about all of this is like, you know, if you do want to do it for free, you can even the harnesses themselves, like you can use open source harnesses like Hermes or even FreeCode. And that way you can get free APIs like HY3 OmniRoot. And then also you can get free harnesses and you can just plug those in. So if you're worried about costs, that's how you can reduce them on an agent operating system. And I actually have a full tutorial on how to run an agent operating system for free, which I'll link to inside the comments there for you. The next question we got was about my digital avatar. So Constitution commented, digital avatar, kudos to the first one I've watched. Looks realistic and smooth. So thank you very much, by the way. This is the real me right now. So it's something we worked on a lot over time and bear in mind, like our old avatar videos didn't look anywhere near this good. What I do every single day is just review my team's videos, check they're actually good, give them feedback and then hold them accountable to that feedback. And over time, you build a really good team with good SAPs and you get better and better. And also the more videos your team create, the better they get and the better the quality of everything you create. So every video we learn something either from the comments and the feedback or from just watching and reviewing the video itself. We also got this question from Library who was asking and basically said like, they've got a Gemini 3.5 Pro one year membership and they still haven't got access to Gemini 3.6 Flash. Maybe that depends on the country you're in, but from my own accounts, I've seen it working fine. So that's really weird. Maybe test it inside Antigravity as well. So if you don't have access to it inside Gemini, it could be worth checking on Antigravity. But honestly, you're not missing out on that much either. Like Gemini 3.6 Flash is definitely better than the old version of Flash, but I think everyone's just waiting for Gemini 3.5 Pro. So as soon as that comes out, it'll be a lot more fun. Google's really like falling behind right now. Another question we got was about our voiceover on these AI avatar videos. So this is the same voiceover that we've been using. Maybe 11 Labs itself is just getting better, but we've been using the same voiceover over, I would say, the last three or four months. I've been improving it, making it better and better. But I still think that AI voice cloning has a long way to go, especially with 11 Labs. I'm actually interested to try and test something like Phish instead. Now, Marikoshi was talking about using OmniRoot with Hermes Agent, but he was saying he still can't do it right. So if you're struggling with OmniRoot, you can actually use other tools. So for example, News Portal has a bunch of free APIs that you can plug into Hermes Agent. So to do that, go to your terminal, then you're going to type in Hermes model, like so. And then from here, you can switch the APIs that you want to use. Now, if you want to get those free APIs with News Portal, just go to News Portal here, log in, subscribe and connect, log in with your account. And you can see I'm on the free subscription. And then we can actually choose the free models who you want to use here. So you can switch between them. And there's a bunch of free models with Hermes News Research. So for example, Ling 3.0 Flash is free with Hermes Agent. And then you can code with Hermes instead of using OmniRoot if you prefer that. Invisible says, it's insane how much content you produce. Yeah, I mean, we do create a lot of content, to be honest. But it's purely because it actually works. You know, like I've always found that when you create more content, you get more attention, as long as you're quality controlling the content along the way. So for me, I have a team that help me with this. They can create the content. I quality control it. I give them feedback. Then we make sure it's actually good and we improve from there. And also, it's so easy because there's so many new things in AI that it's just a fascinating world where there's lots of new updates every day. Bongolo says, why is the extra high reasoning option unavailable for GPT 5.6 Sol in regular chat on my chat GPT business account? It appears in work, but not in chat on both app and browser. Let's have a look at this. So if we go over, I think I've got a business account as well if I check here. In fact, I've deactivated it because I don't think I was using it that much. But if we go to select a model, I've got extra high. So I just checked on this and I actually have extra high, but I don't use the work account anymore. I'm just in the normal account and it seems to work. Not sure why you don't see it, but I would definitely contact support if you haven't already. Kyle Pesh says, I'm a Gemini Notebook Ultra subscriber and I'm facing a serious issue with the slides feature. Initially, custom prompts were not working and now no slides are being generated at all. Both the default slides option and custom prompts fail every time. The generation starts normally, runs for about a minute and then stops without creating a single slide. So if you're not familiar with this already, Gemini Notebooks is basically the new version of Notebook LEM. So you can create a project like so. And he's using Gemini Notebook Ultra, which is what we've got as well. So we're going to test this out now. And if you haven't used Gemini Notebook, it's really powerful. So if we go to a new notebook here and then we'll click on slide deck. Just need to make sure we actually have some sources loaded in. And then you can see this is being created over here. So it seems to be working fine for me. Might be worth you either trying on a different account or actually going over to the app on your phone instead of your browser or try a different browser completely. So three different options there that might work for you if it's still not working today. Also, delete the cache on your Notebook LEM page and that might help to just clear up. We've got a question here, which is, is Hermes Agent different from Hermes OS? So it's quite different. Basically, Hermes Agent is pretty basic and you can only use Hermes Agent directly with it. That makes sense. But if you're using the Agent OS, you can have a lot more customization. So, for example, you can have different agents working together. You could have Claude, Hermes, Anti-Gravity all working together as you can see inside our Agent OS. And then also you can build custom workflows. Like, for example, we have Hermes Oracle and Hermes Astros working together to research our competitors and research trending topics in our niche. And the UI is totally different as well. So what you can also do is you can use this with Paperclip and there's a group chat with Hermes and other agents too. So there's a lot more differentiation to Hermes OS than there is to Hermes Agent. And Hermes OS is just something that we've built and then we give away to AI Profit Boarding members. We've got another question here about Buzz. Now, if you've not seen Buzz, essentially it is kind of like Slack for AI agents. And it's been one of the most interesting things I've seen so far personally. It actually works. Seems to be free, open source. It was created by Jack Dorsey. So it's pretty good. I've got like full video tutorials on it. So I think you could actually get Buzz to control your Agent OS system because it's linked to Claude and Codex, both of which can orchestrate your Agent OS. So if you prefer to control it via the Buzz interface instead, you could definitely do that. You would just ask Claude directly. Bear in mind, like the Claude code and the Codex you use directly inside Buzz is the exact same one you have inside the desktop that can build custom features. So you could actually build custom features into your Agent OS operating system using Claude desktop or using Buzz itself. Now, we actually created a tutorial on ChatCity voice, which I actually really like. Like it's a fun tool. You can see Robert here says, I'm using it. It is brutal. Will get better. But right now, oof, unusable. So I've found similar situations where I definitely couldn't rely on ChatCity voice all day. It is a fun tool. I think it does represent where the future is going with AI tools. But at the same time, it is broken when it comes to computer use particularly. So it really struggles with computer use. And also, it can be a little bit slow to respond. So I definitely think this is a very beta early version. But give it five or ten years. Where are we going to be in the future? That's what scares me. Byron actually said in response to VoiceUpdate, as a tool, I support it that raises privacy consent concerns as well. And I definitely agree. I think it's a scary future. If we're all talking to voice agents and we're just coding away in our offices, that is a very weird and scary world. Where the voice agent knows more about you than anyone else. And also, I think you definitely need to be careful what you tell it and how you conversate with it. Because it's going to be very easy for you to confuse AI and its voice with human. Like your brain doesn't really know the difference. So it's a tool, but it's a tool that has to be used carefully like anything else. Great point, Byron. Hair says one day AI will delete the whole internet. After that, all our modern problems are gone. That is an interesting dystopian future. Who knows what's going to happen? I don't know if deleting the internet will be a good or a bad thing. But I do feel like the world is better with the internet overall. Lenny says no way this guy is impressed by this Minecraft game the AI made. And I think I'm impressed when I see the visuals. But for sure, like when we create these tests with Goldie Bench, are they like 100% production ready? No, but I do like to test them and just see, you know, what the vibe is. And that in itself impresses me. And I do think that Opus 5 was a good model. However, actually nowadays I would say it's a bit too slow. It seems to have slowed down since it first came out. And actually after testing it more, it's very easy to get distracted with Opus 5. It seems to like circle around the problem instead of just being super direct. By the way, if you want to see what we created and the Minecraft game that I was easily impressed by, sadly, then you can check that out on Claude Opus 5 on Goldie Bench. And then if you go over here, you can see all the creations that we've made and you can test it out for yourself. See what you think. Let's have a look at this parachute game. This is not bad. Here we go. By the way, that slide deck definitely works. Just test it out now. Looks great as well. Really, really nice. Frank says, going forward, please say if features a Mac OS only. I think ChatCity voice does work on Windows. Let me just double check for you. And yes, it is available on Windows. So you can use it on Windows if you prefer. We've got another question from Cricket. He says that chat chippity voice consumes credit too fast. I actually didn't run out of credits, but I have done in the past when I've been testing Codex. In general, I tend to find that Codex runs out of credits very quickly, particularly if you're trying to build something half decent with it. I think it's a lot more brutal in terms of tokens that you can use on chat chippity voice versus just using normal tokens versus using like, for example, Claude. Jack says at least it would be useful for Hermes Agent and Blender MCP to work for, you know, people who are not very professional. I absolutely agree. I think the Blender is not available or not usable for most people. It's just too technical. Most people give up or just simply don't have the time to handle it. So it solves a massive problem. If you're not familiar, basically, you can use Hermes Agent with Blender MCP and it can create like really nice 3D mockups. So if you want to create like, for example, product promos or something like that, you can use this. Pullman commented on our video about Ling and said, this is great. My family's half Chinese. I live in China. I am an electric engineer PC nerd. I'm going to check this out. Nothing with some real competition with Americans. The AI race would be very boring without it. I agree. I think a lot of these Chinese models are really amazing, particularly Kimi K3. And also a lot of them are coming out for free. Like, for example, Ling 3.0 Flash is available for free on Open Router. So I think the competition, wherever it comes from, is good for ChatGPT or Anthropic. Imagine a world where there was no competition for Anthropic or ChatGPT. That would be so crazy. Mehmet says there is no Turkish language support on our content. So I think normally the foreign subtitles and captions come after like 24 hours on YouTube. So if it hasn't arrived yet, just check back after 24 hours and normally you'll see them. But yeah, it's a bit slow. So that's basically it for all the questions today, my friends. If you have questions, just leave them on this video and I'll be happy to help you. I'm going to do these daily updates every day, as you can imagine. If you're not sure what's going on here, you know, feel free to check out the video inside the description. And, you know, I really appreciate you watching. If you're wondering what I'm working on right now, I've been testing a lot with Buzz. That's been pretty fun. And then also growing the AI Profit Boardroom community. Right now, a lot of our channels are doing pretty good. One thing I actually worked on recently was just setting up the SEO podcast. So this is the Julian Goldie SEO podcast. I actually created this because one thing I noticed is like on my main channel, I only used to talk about SEO and then AI came along and then I would talk about both. And then it's got to the point now where I think so much of my audience is only interested in AI that every time I create an SEO video, nobody gets. And like it's got to the point now where I can launch a new channel like this and it will get like more views on this SEO video, for example. It did pretty well, 2K views. That gets more views than the SEO content on my channel with like 410,000 subscribers. It's crazy. So it's one of those where I had to kind of like segregate it. And then we also created like a podcast across all our other channels with this. And yeah, it's been a lot of fun doing that. And then also working on the agent OS, that's been a lot of fun. Although I feel like it's got so many features inside at this point, like there's no point adding more because it's just, it's already at its max capacity. And additionally, just creating videos for a podcast channel like this has been fun. So, you know, it's a separate channel, human only content. I kind of wanted to, this was something else I was thinking about was like, I want a channel where it can just be authentic. And then another channel where it's more like AI avatar content and that sort of thing. You sort of like split them so that the people who get the human content, get the human content. The people who want the AI updates and more of them, they can get them on the main channel. And then, you know, everyone is winning. And you've also got separate podcasts for SEO if you've followed me since the SEO days. So really appreciate watching this far. And that is the whole setup. So yeah, comment with your videos. And if you haven't already, check out the AI Profit Boardroom. Link in the comment subscription or go to theairprofitboardroom.com. Hope to see you inside there. This is my AI automation community that helps you learn, save time, grow with AI automation. We've got a community where you post questions. I create a video tutorial for you every day. Inside the classroom, you can get all of our new courses and updates. Inside the calendar, you can jump on four weekly coaching calls, share a screen, ask members questions, etc. And then inside the map, you can meet people in your local area. So hope to see you inside there. Cheers for watching. Bye-bye. Bye-bye.