--- title: 'How to Use Codeex and Claude Code for Free with Omniroot' source: 'https://youtube.com/watch?v=1ohV32u1_Yg' video_id: '1ohV32u1_Yg' date: 2026-07-14 duration_sec: 0 --- # How to Use Codeex and Claude Code for Free with Omniroot > Source: [How to Use Codeex and Claude Code for Free with Omniroot](https://youtube.com/watch?v=1ohV32u1_Yg) ## Summary This video is a live stream from the AI Profit Boardroom community, where the host Julian Goldie answers questions about the Agent Operating System (Agent OS), demonstrates how to use Codeex and Claude Code for free with Omniroot, introduces RTK for token reduction, and showcases AI-powered SEO workflows. The session covers practical setups for reducing token usage, using free APIs, and automating content creation and outreach. ### Key Points - **Agent OS Setup and Configuration** [00:00] — The host explains how to configure CLIs inside Agent OS, hide unused ones, and set up Claude CLI. The installation folder contains section 7 for CLI configurations. - **Using Agent OS for Client Work** [05:00] — For client work, configure memory in Obsidian with separate folders per client, or build custom workflows for each client using Claude. - **Avoiding Slow Agents** [10:00] — Too many rules and skills slow down agents. Use forward/goal mode with a judge to achieve goals autonomously without excessive tokens. - **Local AI and Agent OS on Same Machine** [15:00] — It's better to run both on one machine for easier organization and security. Local models can run on the backend of Agent OS. - **Connecting CLIs to Agent OS** [20:00] — Inside the install folder, section 7 covers CLI configurations. Pick the ones you like and configure them separately. - **Obsidian Knowledge Graph Integration** [25:00] — Use Obsidian CLI skill and agent skills for Obsidian to turn an Obsidian vault into a fully connected knowledge graph for agents. - **Setting Up iMessage in Hermes** [30:00] — Go to Manage in Hermes, scroll to Channels, select Blue Bubbles and iMessage, then configure in four steps. - **Using Free Claude Code with Omniroot** [35:00] — Free Claude Code is an open-source project that can be pointed to Omniroot for free coding. Installation is in section 5 of the install folder. - **Omniroot Overview** [40:00] — Omniroot is a free coder that routes across 90+ free APIs. It can be configured with Claude Code and Codeex for free coding with auto fallback. - **RTK Token Reduction Introduction** [45:00] — RTK is an open-source tool that reduces token usage by 60-90% by filtering out padding and fluff from command outputs. It has 68,000 GitHub stars. - **RTK Test Results** [50:00] — In tests, RTK saved 82.9% tokens on average. For example, a git diff output was reduced from 373,000 characters to 29,000 characters (92% less). - **Token Minimization Playbook** [55:00] — Combine RTK with Ponytail (efficient coding), Headroom (context compression), and Caveman (short replies) to further reduce token usage. - **Free Codeex with Omniroot** [60:00] — Codeex can be routed through Omniroot to use 93 providers with auto fallback, allowing free coding without token limits. - **Free Claude Code with Omniroot Setup** [65:00] — Install Omniroot gateway with one command, then run 'omniroot setup codeex' to route Codeex through free providers. Works with 24 coding agents. - **AI SEO Pipeline** [70:00] — The SEO pipeline uses Google Search Console data to find keywords, creates unique content with case studies, deploys to multiple platforms, and indexes automatically. - **Hermes Custom Workflows** [75:00] — Hermes Astros monitors competitors and trends, Hermes Oracle pulls latest news, Hermes Apollo is a voice-activated agent, and an outreach tool automates lead generation. - **Mixture of Agents** [80:00] — Combining two models (mixture of agents) can outperform a single frontier model like Fable 5, as tested on Goldie Bench. ### Conclusion The video demonstrates how to leverage free APIs, token reduction tools, and custom workflows within Agent OS to build powerful AI automation systems for coding, SEO, and outreach, saving time and money. ## Transcript I don't know. [music] >> [music] >> set up codeex and also pre-clawed [music] code with omni route inside the agent OS. Check they actually work with [music] codeex. Make sure that you can preview what we're creating. [music] create guides on both pre-board code with Omniroot and also Codeex with Omni Routt. [music] [music] Yes, you need to make sure everything works. Update your [music] skills to actually check this works cuz you can see a lot of is broken. >> [music] >> Every time you test a model like this and create something with it, you need to check it works. So you can see above that they don't work. You need to fix that. Update your skills. So you remember that [music] [music] [music] >> [music] [music] >> Heat. Heat. >> [music] >> Test this out. Run [music] your own tests. Create a guide on it. [music] Heat. Heat. [music] [music] >> [music] [music] >> Heat. Heat. [music] >> [music] [music] [music] >> Welcome Jay. Thanks for joining. Yeah, doing great. Thanks. How about you? Sorry about the late kickoff. >> [music] >> Uh, we got a lot of training on that in the air profit boardroom. It's like a whole workflow that my team use. Boom. Shakalaka. Nice one, Jay. There's some good features in the new one. Hey hey hey. If you have a situation where your agents are [music] not delegating to you properly, then what you should do instead is ask Claude or whichever agent you prefer to use. Think up your agents properly so they delegate and then just confirm it works as well. That's usually what I do. [music] Absolutely. If you prefer to use Gemini, just go with that. I think if you're happy with Claude Co, then stick [music] with it. But if you're ready for more advanced stuff and more custom workflows, then I'd go [music] with Agent OS. Agent OS has a lot more custom functionality, [music] is way more powerful, but it depends if you want to go with the advanced or the simple route. How old are you, Julian? I am 36. Would it be possible to have a version of agent OS for each team member where the admin would control and remove skills? I think you could set that up. Yeah. With something like tail scale. So you could add yourself as the admin and then just train your agents so that you can control who has access or not. It would just be like custom functionality you ask Claude to build in. What do you think? Yeah. Yeah, I think you could do that. Uh, just ask Philadelphia. >> [music] >> For each test, can you explain exactly what you did in a way that anyone can understand it? Add a bit more [music] context. Update your skills so you always have more context and explain it in a way where anyone can read the experiment [music] and understand what happened. Hey, hey hey. That's a We mix up the music. [music] Thanks Julian for all the great content. Brilliant job. Thank you very much sir. Appreciate the feedback. We're just uh plugging omniroot [music] into codeex and free claude code. So that should be fun which is a a free coder and a free way to use [music] tools like codeex and claw code. Over here, we're using RTK and testing out how to reduce tokens. And then over here, we're testing out some new benchmarks. I wanted to make Goldie Bench a little bit more interesting. So, we got this little task, for example. and a flight simulator that we're working on. What would you say in your opinion has been the most benefit? beneicial uh with the agent OS. Yeah. So I mean it's more for saving time I would say. So whatever you spend your time on that's what I would automate. So for me for example just saving time by setting up the SEO workflows, video agent workflows, all those systems save me a lot of time and that's really where it helps. So it's all about automating. And then also of course like you know if you can create more content you get more attention. And if you get more attention, uh, then you get more leads. So that's the way that I look at it, but I think a lot of clients would be interested as well. Again, it's set up for them and they don't have the time or the understanding to do it. So it's something [music] that you could offer as well. Hey hey hey. Heat. Heat. All right. Happy dump. Which LM benchmark is the most interesting to me? Probably my own. Uh I I just got kind of tired of looking at [music] other people's benchmarks and not really seeing any. like visual tests where you can see okay what's the actual difference here so for example uh we created this system where we can easily see okay what have we built how does it perform is it any good or not and we did that for each test uh so this is what I like to test is like you know what does it look like visually how does it perform in terms of the physics and then also it's really easy to then compare okay what's the quality of this versus others. I think that's the best way to do it. Just test it yourself. That's what I started doing. And then you can compare each one. So, for example, we have this compare section on Goldie Bench, but we can compare each model versus each other and what they've created and how they perform. So, if we look at the comparisons here, you can see the differences. And I think that's probably the best way. Otherwise, very hard to see. Okay, what's the difference and how do they perform and everything else? All right. >> [clears throat] >> How much have I spent on the API for Fable 5? Uh, nothing so far because we just have Fable 5 inside the subscription. So, I just use it inside the subscription. So, uh, you got until, well, you really got like 24 hours pretty much until that goes away. So definitely make the most of it. I think if you had a choice, I would just go with the native codeex option cuz GPT 5.5 is designed for codeex. You're probably going to get the most out of it right there. Did you actually test and build anything with it? Have you actually made sure Codeex is using Omniroot? Where's the conversation history inside the chat? Why can't we preview what we've built? Let's see. I think it'll be like mid July GPT 5.6 comes out. But yeah, it's good that you made the most out of Fable 5 in between. So, I personally wouldn't use Sonic 5. I would go with Opus 4.8 if you can. Today we're going to be answering some of the latest questions about the agent operating system that we've built, how to use it, how to build them, how to get the most out of them, and what it can do, plus the best ways to configure them. So, we're going to get straight into this. This is based on the questions we get inside the AI profit board community. We're going to get straight into this and start answering some of these questions. So one of the first questions we got was like how to configure each of the CLIs inside the agent OS when you set it up. So basically if you're getting our system and you can set this up yourself if you want to but if you're getting our system there's a folder with the CLI configurations and you just ask whichever agent you're getting to set it up to configure each of those CLI. Now, for most people, they won't use all of those CLI. So, you can hide the ones you don't want inside the toggle in the top left. [snorts] And then from there, you can install the ones you want. Now, for example, if you're getting Claude set up for you, you can just say to Claude, "Hey, set up the Claude CLI inside the agent OS. Check that it works." And then if you have any others to set up, you can go from there as well. So if we have a look at the installation setup Section number seven is the agency inside the install folder. We've got another question here from Brandy, which is, "How can you set up an agent OS and configure it when doing work for clients?" So I would configure that inside your memory system. So for example, if we have a look at the memory inside Obsidian, you would have different projects in different folders for each of the projects you're working on. So for example, when I'm doing work for the profit boardroom, we have a separate folder for that. When we're doing work for our funnels, we have separate uh folders for that too. And so for each product or for each client, you can have separate folders inside your Obsidian memory. And that way you can easily categorize between them. The other option is that you can build custom workflows for each client. So you could ask Claude, okay, I want to set up SEO workflow for client A. And then you just configure it that way. David [laughter] he's added a series of skills to make Claude code autonomous, but it's made his setup super slow. Yeah, I think there's a tendency sometimes for people to give too many rules and too many skills to their AI agents and that makes them too slow. If you want your agent to be autonomous like for example claude code and you want it to work on a project where you just hand it off, you can actually use something called forward/goal mode and you can give it the goal and then it runs autonomously on a loop to achieve that goal and then a judge will actually check the work on every round and make sure it's actually done. So for example, if we go into codeex has the same sort of situation. You can go into goal mode inside codeex. You can give it a goal and then it will loop round on every round until that goal is achieved. And the way the AI knows the goal is achieved is by checking the work with a separate model, a judge, and that makes sure the work is actually done. Otherwise, if you give your agents too many skills, too many configurations, too many rules, they can use up a lot of tokens and also be super slow and just not get anything done. The next question we got was, is it best to have local AI and agent OS running on separate machines? So Scott was saying, you know, I have a computer just running local LM on one machine and should I set up agent OS to run on another separate machine that I do my work and need to keep running 24/7 on both. So for me personally, I would have them both on one machine running locally. So you can have agent OS running on your machine and then the local models running on the back end of the agent OS. And that way it's easier to organize and also it's just running locally so it's easier to configure as well. Also for security I think it's better to have everything set up locally rather than giving access to multiple machines with that sort of access. Blake was asking, you know, what's the best way to connect each CLI or just connect in general to agent OS to bring on Claude, Hermes, notebook, and so on. Are there any pre-built plug and play agents that we can put into the agent OS for various jobs? And if so, where to access and upload them? So, inside the install folder of the agent operating system that we give out, there's a section 7 just for CLI. And you can just configure each one separately. You don't need to add them all in. Just pick the ones that you like. All the configurations and installations are inside the actual install file. This is a pretty interesting share from Tan. So he was talking about turning an Obsidian vault into a fully connected knowledge graph and he's actually used two different Obsidian open-source projects here. So number one is Obsidian CLI skill. So it's an Obsidian skill for claude code and other agents which look super useful. You can see here for example it's got like different ways of organizing it with files, daily notes, search properties and everything else. And then also there's another one here called agent skills for obsidian. you can basically teach your agent to use CLI and open formats including like markdown and everything else. So, it's an interesting way to just customize Obsidian more so that you your agents understand better how to use Obsidian. And if you're wondering what Obsidian is, Obsidian is just a memory system that you can use to give all your agents context so they understand you, your goals, your business, your voice, and everything else. We got other people who are sharing their setups for the agent OS as well, which looks great. So, the great thing about this community is we all share and we learn and we grow together to help each other as much as we can. Actually, what's pretty interesting here is, you know, for example, like Jacob, he's posted that you just set up my agent OS in about 10 minutes. And he said, "That's a strange feeling because I've spent the last week on my own version, 12 hours a day, and in 10 minutes, he set up something that took him like 12 hours a week." So that's the power of this stuff is like it can really save you a lot of time. It's super powerful and really easy to use. Now we got a question from Scott which is how to use iMessage within Hermes how to set up etc. So, as far as I'm aware, you can go to manage inside Hermes here. Then you can go to channels and then inside channels you can select blue bubbles and iMessage and just configure it. So it just works in four simple steps. Manage inside Hermes, scroll down to channels, click blue bubbles, and then blue bubbles, and then click configure. And then once you're set up, you're good to go. also what's pretty interesting here is like people are setting up Agent OS on multiple devices which is really cool as well. So he can use it on his desktop but also on his laptop as well. So great share from Paul on how to set that up here. So Ronald has also set his agent OS up and he was asking like how do you use free claude code and where does it live? So there's a there's a markdown for the installation of free clawed code. It's in section five of the install folder. And yeah, you do want to get your agents to configure it for you. But once it's done, it's pretty easy to set up. So you can see that we can run pre-clawed code over here, we can see what we've built with it. We've got the workspace over here too with everything that we've created. So here's a landing page that we created earlier using free claude code and you can also configure this with any model. So that could be a local model or for example it could even be like omniroot or something like that. It's based on an open-source project. This one right here. this pretty interesting. So, one question we had was about omniroot which we've now actually configured. Now, this is a free coder that automatically roots the basically across 90 free APIs. So you can code for free. So you can use it inside the omniroot section of the agent OS. Again, it's an open source project where you can just keep coding even on free APIs and you get preview and all your workspace and everything else. Now you can configure this with free code and also codeex. The way that I personally set it up was I went into Claude and I just gave it the documentation for Omniroot and said, "Can you set this up with pre-clawed code and codeex as well?" And it just went off and set that up for me. But I think you don't need to 100%. It's just like some people prefer the harness of Claude code. And so that's why I configure it like that just to show people how to do it. and that is it. We have answered all the questions inside the AI profit boardroom. If you want to ask me questions like this, just go inside the AR profit boardroom, post your questions and I answer them with video tutorials like you've seen today every single day and every single question gets answered. Also, the whole community helps each other as well. So, when people have questions, we all sort of chip in and help each other as much as we can. And also inside the classroom, you can get access to our agent operating system. So you could build this yourself, but like I was showing for example with Jacob, like sometimes it can take like 12 hours to set up something like this. I personally code this out for like 3 to four hours a day. So if you want to get my system that's updated with all the new changes, all the integrations, etc. You can get that right here inside the classroom of the AI profitable boardroom link in the comments description or just go to the profitable.com. And this is my community for helping you learn, grow, and scale with AI automation. So inside the community you can ask questions, get help and support in real time. Inside the classroom you get access to all my best trainings on this sort of stuff. We have new daily updates. So for example, yesterday we actually set up a full training on how to get free APIs. So if you want to get free APIs, uh there's actually over 130 of them inside Omniroot and we show you exactly how to do that inside this training right here. And then if you want to plug this into your agent OS, you can grab over here. We also do weekly coaching calls so you can get help and support in real time. Share your screen and inside the map you can meet people locally who are building with AI agents like you. So feel free to get that link in the comments description or just go to the apiprofitbomb.com. Thanks for watching. Thank you very much. Appreciate it. Always happy to help. When do you think GPT 5.6 is releasing? I would expect it to come out mid July. How can I get APIs? So, you can use omniroot. Omniroot is pretty good for free coding. Yeah, I saw I saw your I I keep seeing your questions about this. So, let's have a look at this. Does your Claude call you test? Um, where was it calling me test? I wasn't even aware of that. Do you have an example or where you saw it etc. If you let me know where you saw it, I'll I'll answer and just check it out for you. Let's have a look. What is my name? Where does it call me? Good. Where does it say good afternoon test? On which screen? If you let me know where, I'll take a look for you. How can you get I don't think you can, mate. I don't think you can. Yeah. So, if I ask what is my name? It knows I'm Julian Goldie. So, I don't think I'm getting that issue. Go to claw.ai. So it says, "Good morning, Julian." Yeah, I don't think you can get a free API for Claude, but if you want free APIs, you can get them with Omniroot. And we created a couple of tutorials yesterday about it as well, how it works etc. Yeah. Is that my videos or is it like my teams or which ones? It could be my team could have been uh on one of my teams examples. Thank you. That [music] is pretty wild. It looks really good. >> [music] [music] [music] >> Back down. Heat. Heat. >> [music] [music] [music] [music] [music] >> Wow. What on earth? [music] Heat. Heat. [music] >> [music] [music] [music] [music] [music] >> So today we've been testing out RTK which is a really powerful way to reduce the amount of tokens you use. In fact on our own test and I'll show you proof in a minute. We actually saved 82.9% in tokens. Now, the readme of RTK actually claims 60 to 90% savings in tokens. And this is just a really powerful way to get more out of your models without paying more because you can reduce the amount of tokens you use and be more efficient, which means you can get more out of it, especially if you're using something like Fable 5 on the subscription. There'll be models in the future that you probably need this for as well. I'm also going to show show you how to combine this with other open source projects to reduce the amount of tokens you use even more in terms of uh being more efficient. So, this is a really powerful system and it's got 68,000 GitHub stars. It's trending right now. Pretty fast to use. And if you're wondering, okay, what does this do in plain English? Because you'll probably read this and wonder, okay, what what on earth does this mean? So it basically means that you have a filter that sits in the middle between you and your AI agent. So you still get the same back, you still get the same outputs and they're just as good, but all the padding, all the fluff, about 80% of it actually gets removed in between. And so what this means essentially is that you just reduce the amount of tokens to use depending on what you're trying to do. And you can see a bunch of tests that they've run over here. Average saving was 80%. And it's pretty quick and easy to install. It works on Linux and Mac OS. So you can install it in one single command. And then once you've done that, you can start using it. Now this could work for example inside Claude. It could work inside all your favorite coding agents. And it's pretty simple and easy to use. So, I'll give you an example. On one test that we did, we had 82.9% less text for the AI to read across all of our tests. And we ran loads of different stuff, which I'll show you in a second. Also, there's no real delay. So, it was like a 14 millisecond delay between using the filter and not using it. So, an extra 14 milliseconds, which is not too bad. That's not going to slow you down when you're using it as well. It's basically like a civ for your computer's replies and that just removes all the fluff in between. So let's say for example, you ask Claude to fix a bug. Well, this would be super useful. Or if Claude asks your computer a question, those sort of things will reduce the amount of tokens you use massively. So here here's how it will work step by step. So if you asked Claude to fix a bug or on something you're building, that could be a website, tool, app, game, whatever, then Claude would ask your computer a question to check that it works. Now without RTK, my computer would reply with like 373,000 characters according to our tests. And we we got Fable 5 to test this all out in terms of the facts and everything else. And so Claude has to read all of it and that text sits in its memory being read. That means that you use a lot more tokens on every message after that. If you use RTK, then a small rule swap swaps a command to RTK get diff instead of git diff before it runs. And that means Claude doesn't really know, but it typed the same thing. voice types. And so this means that it runs the exact same G command, but it strips out all the padding. So all the repeated headers, all the unchanged lines, all the decoration, and what's left is the actual change. So in the end, we would normally get 373,000 characters back, but with this, we got 29,000 characters back. And so it's 92% less for Claude to read. So Claude then reads the short version, fixes the bug exactly the same before. you know, you get the same result, but it's using a fifth of the tokens hour after hour, command after command. And if something actually goes wrong, then RTK saves a complete unshortened reply to the file and prints it where it is. So you don't ever actually lose anything if you need to revert it back. And also, the good thing about this is it doesn't change how you work, doesn't change how the AI works. The replies just get shorter. So here's an example of how it works. So normally you have your agent, it'll run like gear diff. Then the shell answers honestly and you get your context back like 373,000 tokens. With your agent, it would run the same, but it gets filtered through RTK. The same command runs in the shell and then you get much lower context back. you get 92% less context back. And so when it filters and truncates everything that's doing and reduces to duplicates, you just use a lot less tokens. Now, the way that we've tested it is we used it on our agent OS dashboard. We ran every test with one command, but we ran it twice. One the normal way and one through the RTK in the same folder on the same machine with the same minute. Um, we measured the size of what came back. So, every command answers in text and we counted the characters of both answers because that text is exactly what an AI has to read and exactly what you're usually paying for in terms of tokens. Now, it's pretty easy to install. So, it's just like one line to install this and then you get the latest RTK version and it's a really small program. It's like 6.6 6 megabytes. And it knows how to shorten the replies of 100 everyday commands that Claude would normally use. So we tested it for example with uh auto rewrite hook with 14 agents including Hermes a bunch of other stuff like this but everything seemed to reduce the amount of tokens that we were using. Now you might say if you're filtering stuff you know if you're filtering the output doesn't that mean you lose information? But you actually don't lose information. It just reduces the amount of tokens that your agent would use. Now, also what's interesting about this is if you want to reduce the tokens even more, you could stack this in with the rest of our token minimization playbook. So, the things that we're using right now is ponytail and that's basically way more efficient in coding. We have RTK which shrinks what commands return and then you can also use headroom which compresses the whole context. Now if you're not familiar with each of those RTK is the open source project for reducing LLM token consumption. Then you have ponytail which basically makes your AI agent think like a lazy senior developer to be super efficient. And additionally, you have headroom which compresses the amount of tokens you use as well. Now, on top of that, you can also use something called Caveman. And these are all free open source projects. They're all pretty popular. I mean, for example, Caveman has 85,000 GitHub stars. And this basically means that when you're using Caveman, it will reply to you with less output. And so each one of these playbooks helps you reduce the amount of tokens you use. So for example, if you're using caveman, if we say for example, what is my name? It just says Julian Goldie. And it's like caveman mode is now live in the session, which means it's going to reply to me in a very short brief way to reduce the amount of output tokens he uses. So if you have a look before and after with a normal agent, it would give you a long response like that with a lot of fluff. With caveman, it would be much shorter. With ponytail, it reduces the amount of code it creates. And then with RTK, this just filters the amount of tokens it uses as well. So these are all like good ways to reduce the amount of tokens you use to be way more efficient. and then you can plug them all into your agent to reduce the amount of tokens you're using. And this is actually part of something we call the token minimizer playbook, especially when you're using Fable 5. This is a great way to get more out of Fable 5. So, for example, this week we've been testing Fable 5 relentlessly with loads of different tests, 47 different tasks, but we use that all on the subscription of Claude because we were minimizing the amount of tokens we use. We created some like pretty crazy stuff here as you can see, like full real world games as you can see. Um, so if you want to get more out of your plans, out your subscriptions. And bear in mind, like the bigger the context window of each of these models gets, the more tokens they're going to use. So when you can be more efficient with these playbooks, just going to help you save a lot more. So that's basically it. They're free to try as well, so you can test them out, see what you think. If you want to get a full machine, I mean this is part of the agent OS setup. So if you wonder what the agent OS is, it's an agent operating system where you can plug in claude open claw, you can add Hermes, you can add memory systems inside there. And a lot of people say like want this use a lot of tokens, but we use it with our existing CLI plans and free APIs simply because we have smarter playbooks in place to reduce the amount of tokens we use across everything that we do. So if you want to get that, it's inside my AR profit bomb community link in the comments description or go to the arrum.com. Inside the community, you can ask questions and I answer them personally with video tutorial every single day. Inside the classroom, you can get access to all my best trainings including new daily updates. We've got the agent OS system here. We update it daily. You get video tutorial for it and also a zip file to install it. And then we add new daily guides based on what's actually used for as you can see right here. Uh we also have four weekly coaching calls. You can get help and support real time. And inside the map, you can meet people in your local area who are building with AI agents like you. So feel free to get that link in the comments description or just go to the aiprofitwarm.com. Thanks for watching. Thank you. Yeah. Well, I mean that's pretty good. Like if you're only on 81% and you got one more day, you've used it very efficiently there. Hey Jose, thanks for joining. [music] I would look at, okay, what do you need to automate? What do you spend your time on? And then reverse engineer from there. But if you want to see some examples of what we've built with it, you can see a bunch of examples right here on Goldiebench. And it's built some pretty cool stuff. Like some pretty amazing stuff. I mean, even like the videos it's generated super nice. [music] Looks really good, actually. [music] But yeah, if you [music] want inspiration, check out Goldie Bench or check out the Agent OS as well. Got loads of cool stuff in there that's built with Fable 5. What are you? [music] [music] >> [music] >> Daddy, daddy. >> [music] [music] [music] [music] [music] >> Back in a minute. >> [music] [music] >> Hey, hey hey. >> [music] >> Noo. Nah. >> [music] >> Where are you? [music] >> [music] [music] [music] [music] >> Heat. Heat. >> [music] [music] >> Nat. [music] Nah. >> [music] [music] [music] >> Hey, [music] hey, hey. [music] Hey hey hey. [music] >> [music] [music] [music] >> Daddy, daddy. >> [music] >> Hey, [music] hey hey. [music] >> [music] [music] >> D. >> [music] [music] >> Daddy, you >> [music] [music] [music] [music] [music] [music] [music] >> Today I want to show you how to use codeex for free. And this is a super powerful system you can basically use to code [music] across like 93 different coding tools with one single tool and then you can plug it into Claude Codeex. If you're not sure what that means, let me just show you. [music] So this is Codeex and what we've done is we've rooted it through omnioot which means we've got 93 [music] providers plugged into codeex and we don't need to subscribe to codeex to use it. So for example here if we say okay build out a beautiful landing page for an SEO agency we can actually code for free using this powerful system and we can also see the things that [music] we've coded out previously. So this is our agent OS system. We've got codeex inside there and then we've used [music] omniroot inside codeex to build and automate whatever we want. So you can see that's working right now. And then if we switch over to this [music] tab we can actually see the things that we've previously created. So you can see for example here we actually created a demo as you can see here and then we've got the full HTML page here for another test that we did and so we can build and code whatever we want and it actually looks pretty good. So it's using the harness of codeex but it's free to use because we've got Omniroot [music] plugged into that system. So today I'm going to walk you through exactly how it works step by step with what I call the free codeex engine. So it's open as codeex coding agent but it's rooted through omni room [music] across 93 providers with auto fall back when it runs dry. So like for example if one free API doesn't work no problem you can switch to a different one and so codeex [music] is still creating for you but the brain of it is with omniroot which is pretty amazing stuff. So if you're not familiar with omniroot [music] it's just dropped recently. It's already got 12,000 stars on GitHub. I've already tested out inside [music] Omniroot directly. So, for example, we've got Omniroot as a coding tool built in. But if you want to use the full harness of Codeex with Omni Route, you can. And then also, if we have a look over here, [music] you can see the finished landing page that we created here. The other cool thing I like about this is when you build with this stuff, I mean, it looks pretty sleek. It looks nice. The colors nice. It's simple. It's not going to be like Fable 5 level, but at the same time, like the landing page is not too bad at all, considering that was a super basic [music] prompt. And that's because we're using the harness of Codeex to get better outputs. The other thing I would say here is like it's it's fairly quick. I mean, it only took [music] about what 60 to 90 seconds to create that. And everything I've built with it so far has looked really good. So, you can also [music] see the preview here as well. when you're building out inside the chat of codeex we can see the preview and this also links inside our agent operating system you might also say [music] why would you use it with an Asian operating system because then you can plug in the memory the context you can take what you've created with codec you can plug it into cord [music] code you could for example design a landing page but then plug in a video from the video agent inside the agent OS so it all links together it all syncs together and also like [music] I don't need to open up codecs and then claude separately they all work inside one beautiful system And we can all do [music] we can do it for free. Like that's pretty amazing to live in a world like that. So Omniroot is super powerful, super fun to use, easy to use as well. 93 providers in there. It's never failed on me when I've tested it as well. And you can basically just code for free unlimited. Now the the other thing I would say here is like when you're using for example free models with codeex, you can run them locally. For example, like you can go over to Alama and you could get a free model like Laguna XS 2.1 and you could plug that locally into something like Codeex, [music] but from what I've seen, the outputs are much nicer from these APIs that it's using rather than just using the local models directly. It's also a lot faster and if you're using local models with something like Codex or free cord code, the problem is that quite often it it uses up a lot of power on [music] your computer, right? So it's much slower when it's responding with this setup. It doesn't seem to slow anything down at all and we just get really [music] nice outputs as you can see here. So how does this work step by step? So essentially what we have is we [music] have a codeex that goes through omni route. It picks and falls over. If uh one provider doesn't work, you get 93 providers and then you get everything written. So it's [music] the same agent. There's no new tool to learn. It's a real codec. It plans, it writes files, it runs commands on its own exactly as before. [music] It's only the back end that's changing. And also for me personally when I've used GT5.5 one of the reasons I stopped using [music] codeex itself was because it just ran out of tokens really quickly like even on the paid subscription. So that's why I think this is might even be a [music] better alternative in some ways then you can use the free pool. So instead of the open eye meter codeex roots to [music] 93 providers 11 of them are actually permanently free as well which is pretty crazy as auto fallback. So agents [music] can run out of tokens or they get rate limited particularly free ones and so this way it just silently hops onto the next one which means like if you're in the middle of building something [music] and just break in the conversation it keeps going it keeps building. Let's run another test here. So for example if we go inside the chat [music] we can ask codeex again. So, we can say, okay, build out a beautiful [music] to-do list [music] and we'll test it out. Now, again, like you don't need to run this locally. It's all cloud-based. So the free APIs are cloud-based. Codeex is super lightweight to run and with this system you don't need to use a lot of memory. So you could just use this on like a basic laptop. You don't need any fancy setup. And then also you get the previews as well when you when you're creating with [music] it as well. The thing I like about this is like usually if I'm coding within codeex it it gets a bit tricky because you can't find what you've created previously. Whereas with this system we've got the workspace and we can just come back to whatever we've built whenever we need to. And I think that's a better way to do it. So, if you look at the old way versus the new way, the old way like every [music] codeex run is on the open eye meter. Agents are chatty. So, a long build ranks up fast. By the way, that's a to-do list that we just built and it actually works. It looks really cool. You hesitate to use it too much and you're kind of locked into one provider as well. So, for example, if you're [music] using open eye, then you're locked into open eyes provider. With this setup, you got the same codec, but it's pointed at [music] one local gateway with three providers built in, auto fallback if it fails, and you could just have this building [music] all day for you pretty much. You also might say, well, I already pay for chat [music] GBT. Why would I use this? I mean, this can run alongside. So you could have codeex for example as the app locally and then you could have the agent [music] or s codeex here plugged into omniroot and that way if codeex runs out of tokens no problem you can switch to the next one and with omniroot it's just ready to go straight out the box. So [music] you can install it in like one command you configure it like so and then from there you're good to go. like it's it's very very easy to set up literally for setting this up. You just install the gateway. One command spins [music] up omni root like this and then one command rise free profile like that. So you do omni root setup codeex [music] and then it's like one click away inside your agent OS. Now you also might say okay I'm not technical. [music] This sounds advanced but literally it's just three commands. Simple as that. In fact, two commands, this [music] one and this one, and then Codeex just routes to multiple free providers from there. And the great thing about this is, especially if you're using the agent OS, like you got lots of CLI in there, you're probably doing lots of different tasks at once, you got loads of different chats and conversations and agents plugged in. And so, the great thing about all of this is that if you can minimize the amount of tokens you use, or if you can use free providers, then you can use it even more without having to worry about tokens. So that's basically the setup. That's how to use it. It creates pretty awesome stuff. [music] Uh some people think that if you want codeex, you have to pay for open API per token, but you don't with this setup cuz you're running the same codec, but only the back end changes. Other people might say, well, rewiring a coding [music] agent provider is advanced stuff, but it's just one or two commands as you've seen today. And other people say, well, free AI it always gets rate limited or it stops working halfway in between. But the great thing about this setup is the auto fallback fixes that. So when one provider [music] caps out, the run moves to the next automatically and it survives to the finish. You can also, if you want to minimize the amount of tokens you can use, we have a token minimization playbook. And there's a lot of cool ways [music] you can reduce the amount of tokens you use. So, for example, you can use open source free projects like Caveman, Headroom, Ponytail for coding, and you can also use RTK. And these are all ways that you can combine multiple different open source projects together to reduce the amount of tokens you use so that you [music] don't miss out on the plans you're trying to build out. >> [music] >> What would be interesting as well if you run this with goal mode? I haven't tested that out, but like if it's got auto fall back, and you can keep looping around autonomously. [music] So with goal mode, if you're not sure sure about what it is, basically you can give codeex a goal. It loops around, [music] has loads of turns to complete the task. A judge judges if the task is completed. And if you ran that with Omniroot, [music] I wonder if it could just keep going for hours with the auto fallback. I haven't tested that out, but that would be quite interesting to see as well. And then you can see everything that we've done and what we've built over here as well. So, it's pretty cool. We can also [music] open that in a new tab. We can save it. We can copy the [music] code. We can switch to preview. Um, we can test out everything we've built. It's pretty cool. So that's basically it. Now the free codec engine is just one tab. The agent OS is a whole room. If you want the free codeex coder setup inside our agent operating system, [music] we've got the full setup inside the proper volume. So you get a zip file to install it and you'll learn how to use codeex for free. We've also got free call code plugged into that. We've got all of our main CLIs and agents plugged in. There's 3,900 members and myself inside there and you can ask me any questions, get help and support whatever you want to. So, if you want to get that link in the comments description or just go to the aprofit.com inside [music] the community, you can ask questions and I personally answer these questions on a daily basis with a video tutorial. And then inside the class, you can get access to all of my best training. So, if you're a beginner, you can watch the beginner to expert course over here. [music] If you want the agent on our system, you can get over here with a video tutorial. You can see when it was last updated and you can also [music] get the zip file to install it. And then we also add new daily tutorials based on what's actually useful. So we've got a full guide on Omniroot, how to [music] use it, how to get the most out of it. And also every time something new and useful comes out, we share inside here. Inside the calendar, you can jump a week coaching [music] calls, get help and support in real time. And inside the map, you can meet people in your local area who are building with AI agents like you. So feel free to get that link in the comments description or just go [music] to the aiprofitborn.com Steven. Yeah. So I I do the live streams like this simply because for me those calls are way outside my time [music] zone so I can't make them. But I do like three or four hour sessions a day like you see today. And also I'm not [music] really a coach. So that's why we've we've hired a coach who's like really good at helping you. But if you need to [music] connect with me personally, you can just direct message me anytime. And I create video tutorials for everyone inside the community every day as well. [music] [music] >> [music] [music] [music] >> Heat. Hey. Hey. Hey. >> [music] >> Hey, I'm trying to give my Hermes agent a voice. any recommendations? Yeah, I would I would use the Hermes Apollo system that we've got set up inside the agent OS here. So, if you go to Apollo over here, we can basically speak to Hermes as you can see. So, [music] Hermes, how day going? So [music] you can see an example right there and we can chat with this in real time. So if you want that full setup for the uh voice agent that you can speak to live with Hermes agent, you've got it ready here. Um and you can use [music] it whenever you want. Let me show you another example. Hey, how you doing? [music] Can you just can you just open up Google for me? Boom. [music] There we go. Works really nicely. So feel free to get that. >> [music] [music] [music] >> Yeah, for that sort of system, I would I would create it with the agent OS, right? So, plug it in as a custom workflow, describe to Claude exactly what you want and ask it to build it into your agent OS and then you can sort of go from there. [music] So, for example, with Hermes, I asked it to build an outreach tool and have a section where I can, you know, type the type of leads I want, send the emails, etc. And we just built it inside the HOS. And it was just a case of like me describing what I want, asking Claude to build it inside there, and then it went from there. So, that's what I'd recommend for you. Just, you know, whatever idea you have, get Claude to build it, and build it in a place where you're going to come back to it. So for example the [music] agent OS I come back to every single day and so that's why we built it in here the same for example with the mixture of expert section here. So [music] with this whole system that we've built we basically just asked Claude to create a mixture of experts tool [music] inside the agent OS and it built it. So whatever you want to build like clock and just build it [music] for you. You just have to describe the idea that you want [music] Daddy, [music] daddy. [music] >> [music] [music] [music] [music] [music] >> Yeah. So, if you have like any I would just get better internet then. It sounds like that's a constraint. So, if you I mean, if the internet signal is not good, just just get a better internet or or do the work from somewhere else [music] or get someone else to build it for you if you if you don't have time or if you don't have good internet or whatever. Or you could get like a really good local setup that just works offline and can build [music] out like a DGX Spark or something like that. Today I'm going to show you how to use Claude Code free forever with [music] a new setup that's just dropped from Omniroot and Omniroot basically allows you to code with over 231 free providers with one single endpoint and then you can plug that into something like free [music] core code or into Codeex and you can code for free forever with this agent harness. So for example, if we go inside [music] here, this is our Asian OS and we've already plugged free claw code into it. And free claw code is an open source project that you can point to any frame. So we've plugged it and pointed it at the omniroot frame. So if we go inside here and we're like, okay, build out an SEO agency [music] website like so, it will start to build out as you can imagine. And then if we want to see what we've built, [music] so for example, we can wait for that to start working over here. And then if you want to see what we've built previously, we can just go into free call of code, go into the workspace over here, and we can see all the stuff that we built, right? And it can build some [music] pretty nice stuff. Like for example, this website we created with free code and omnirute. Here's another example. [music] Here's another example of an app we built. And this is pretty easy and simple to do. In fact, we've got loads of stuff that we built out here with free [music] code. Now, out of everything that I've tested, it's pretty interesting because out of everything that I've tested and I've tried like many different free APIs inside free claude code and also local models and I would say omnirute by far has been like the nicest to create great stuff. So, whilst that's working in the background, let me show you another example. [music] We actually plug this into codeex over here and we can ask codeex to work in the same way. It's using OmniRoot. And then if we go to our workspace here, we can see everything that we built for [music] free with Omniroot as well using codecs, right? And it looks super nice. Like some of this stuff is [music] really nice. Now, how does this work? Well, basically this is the open source project, the GitHub. And essentially it means [music] that you can use the Kodo directly. It's a free AI gateway. Connects to like 237 providers. 90 of them are free and it works through one [music] endpoint. Now you can plug cord code or codeex or cursor or client copilot even anti-gravity into free core code. The great thing about that is like a lot of people are worried about rate limits on free APIs or for example using [music] local and sometimes local is too slow or it's not powerful enough or for example you just you don't get the sort of frontier outputs you want from something you're normally used to. The other cool thing about this is when it's being used, it combines RTK [music] and Caveman, which are two ways to compress the tokens you use to save like 15 [music] to 95% tokens. And additionally, when you're coding out with this stuff, the other great thing about it is that it automatically [music] falls back. So, for example, if one API stops working or isn't available or is rate limited, no problem. It can switch to the next [music] one over and over again. And so this is a pretty powerful way to use cla code for [music] free with this setup. And so I've got like the free chord code engine. And so you're using like the actual cord code, but it's rooted through omniroot across 93 [music] providers. And you got auto fallback uh fallback when one runs [music] right. Also, you don't need to subscribe to claude to use this. And with you can embed it inside your favorite [music] operating system. So for me for example we have the agent operating system for claude openclaw hermes everything else and then we have [music] a separate section for free claude code for codeex and also for omni route. Now you can also code directly with omniroot here as you can see and what this allows you to do is you can use the same system but you can basically see a preview of what you've built. Then you've got the saved stuff that you've created. So for example here we've got all the mini apps that we've built. One thing I will say is you're not going to get like Fable 5 level outputs with this, but if you just want something to code out some quick ideas or build some mini apps or for example create like a few landing pages and blogs, etc., that's perfect for this. Like that's actually going to work quite nicely. [music] And the great thing about this is you don't hit limits. You save 95% of your tokens. [music] It's $0 to start. Every tool actually works. It works with over 24 coding agents which is pretty wild. There's one endpoint and it's production grade as well. So like before a lot [music] of people love to cord code but every session is limited by the number of tokens you have available. And quite often particularly if you're using claude for a lot of big projects you just run out halfway [music] through or you get token limited. So for example, we go directly to Claude over here. You can see it says approaching session usage limit. We are limited in [music] terms of what we can build because even though we're subscribed, we have a limited number of tokens. With this setup, we can use the exact same claw CLI, but it runs through 93 [music] providers. And if one runs dry, it hops over to the next. So we never get cut off. We can build all day [music] and we don't need to worry about the token limits. And it's a tab inside the agent OS. So if I [music] type it writes files and I can preview them live. So you can see for example it's actually [music] built out the full size structure for what we created. And then if we actually want to see, you know, what it's building so far, we can see that over here. [music] So you can see when this was updated, like 2 minutes ago, 2 minutes ago, we've got the outputs inside the chat and it's just building and working in the background here. And then over here, we can actually preview what it's done so far. So for example, if we click on this page, it's already built out the services page, which looks super nice and it's easy to use. Then we can see the source. We can open that up in a new tab if we want to as well. And this is all coded out locally. So what that means is like you're using free APIs in the cloud but the actual outputs the files that you build the HTML the code etc is coded locally so you can come back to it whenever you want inside your workspace. So if you're like ah where did that [music] page go? Well you can see it now and you can see that it's actually building out the full blog as well which is pretty cool. So it's building out like [music] a full website organized it actually looks quite good. We can go back and forth inside the chat as well if we want to. And the other benefit of this is like because it's linked to our memory setup. So it's using our full Obsidian memory [music] knowledge center here, you can use that to create better outputs inside the chat here. Right? So if you have ideas on like how you want the voice to be or [music] how you want the page to look, you can actually add style guidelines inside your memory and then free claw code can link to that and use that to actually write better [music] outputs. So that's pretty powerful, pretty easy [music] stuff. And then the way this works, it works in five [music] layers. So you got claw code CLI that's running through omni room, you get 93 providers, and then you get your build and that all works inside one tab inside [music] the HOS. So you get the prompt section, a workspace, and a live preview. Now, you also might say, "Okay, well, three models don't match the real Claude." [music] What I would say with that is like again, you're not going to get Fable 5 level outputs with this, but no other model [music] matches that anyway. But what you will get is something you can actually use that's actually useful that actually looks good. You can build out [music] four projects and it's easy to set up as well. So, it's literally coded this whole website [music] whilst I'm talking to you. And if you look at the page, like it it looks pretty good. And you can always improve this as [music] well. So you can go back and forth inside the chat. You can tell it what you want to improve, what you want to change, etc. But it's all set up and ready to go. And that's a full website fully [music] automated as one project with free clawed code. So free models [music] can still create really good stuff. And the other difference here is like when you're using this, you don't need Frontier models for everything. So, if you're just coding out a simple landing page or a website or a blog post, you don't need Frontier models for that. You can just use free APIs like this. Now, also there's a bunch of settings that you can change to. So, for example, with Omniroot, you can optimize depending on what you're doing. Like, if you just need fast outputs, you can change to auto slfast [music] as the model ID. If you want really good coding outputs, then you can switch to auto/coding and it focuses on quality [music] coding generators. So depending on what you want, you can switch, you can even switch to offline as well, which is pretty cool. And then you've got a bunch of different routing strategies. So depending on [music] how you want the models to be used and which models you want, you can set that up as well. And so if you look at the old way, you would be using a lot more tokens and you'd be worried about [music] everything you're creating. With the new way, it's like I can build out a full website. I don't need to worry cuz it's on free APIs anyway. [music] And if I do need to change anything, we can do that later. You can also see here, for example, that you can run free claw code in about 5 minutes. So how you do this is use this one command. Then you're going to switch [music] it to cla code and then it's just going to be one tab inside our agent OS. Now some people say you know [music] I'm not technical. This sounds difficult to set up. So what you can actually do is you can just use these terminal two terminal commands. Pretty simple and easy. Well the other option that you have is that you can actually go into claude and you [music] can give it the details of the GitHub and ask it to set that up for you with Claude code. [music] So if you're not sure how to configure it, what I did is just say right here's a GitHub set this up in cloud code inside our agent OS. I actually went off and did it and it was super smooth. So that's basically how it works and then cla code will route [music] to multiple different free providers. You don't need to be technical to set up and also if you're using agent OS and you've got loads of CLI in there, you can build loads of cool stuff, but you don't need to worry [music] about tokens cuz you're using free APIs. And then for like the more Frontier level stuff, of course, then you can switch to normal [music] core code. So, thanks very much for watching. If you are wondering like is this technical setup, etc. We've got so many people getting amazing results with the agent OS. As you can see, we actually got over 196 pages of people winning, learning, and growing with our agent operating system. So if you want to get that that's inside the AI profit warning link in the comments description or just go to the aiprofit.com and inside there you get the full agent operating system that you can see. So you get free claude code the workspace you get omniroot that we've plugged in right there. We also [music] have for example codeex plugged in. We've got claude ready to go whenever we need it. We could also use this inside Hermes as well. And then we've got paperclip to orchestrate [music] all of our agents group chat with agents pipeline for going from idea to implementation really quickly. And then also inside Hermes, we have the Oracle, we have Astros [music] for finding the latest training topics, and we also have Apollo, which is a voice activated version of Hermes as well. So if you want to get all of that, it's inside the arr link in the comments description or just go to [music] the arr.com. Inside the community, you can ask questions, get help and support in real time. I personally answer every single question. I answer it with a new video [music] tutorial each day. And then inside the classroom, you get access to all of our best training. So if you're a complete beginner, you can go from beginner to [music] expert over here. If you want to get our new daily updates, you can get that over here. And we also have the full agent OS system. You can get a video tutorial for that. You can see when it was last [music] updated, you get the zip file. And we also add new daily tutorials based on what's actually working. So, we have a full [music] tutorial on how to use Omniroot and how to implement it step by step. And every time something new and useful comes out like this, I test it and then I show you how to use it. Inside the calendar, you can jump on weekly coaching calls, get help and support your time. Inside the map, you can meet people in your local area who are building with AI agents like you. [music] And that's all available inside the alpha boardroom link in the comments description or just go to the AI offerboarding.com. Where do I find all this stuff? Do you have links somewhere? Yeah, go to the airboarding. What about sub agents in Hermes? That's already set up now. So there's a new [music] update and a new tool inside Hermes called delegate task and it can automatically set up sub agents for you. So the great thing about that is that you don't [music] need to uh run sub agents manually yourself. However, if you do want to have sub agents and you want to set up set them up yourself, then we do that with different Hermes profiles as you can see. So we can set up a different Hermes profile depending on the configuration. And then also inside the camb board here, you can have multiple agents working together. So if you want to build out like teams of agents, you can do that with the camb or you can do that with Hermes and separate profiles. Or you could actually set up paperclip and then have a full team of Hermes agents like we've got right [music] here. So there's many ways to orchestrate agents, but if you just want to use sub agents automatically, it's already built in. [music] >> [music] [music] >> Are you still using the Hermes workspace as part of your OS? So, I don't use Hermes Workspace directly [music] just cuz it was a little bit buggy. And that's why we actually built the agent OS cuz the other problem with Hermes Workspace is as cool as it is and is a great [music] project. The problem is you can't use it with multiple different CLIs and other agents. And so [music] what we did instead is we created this system where we can have claude, open claw, Hermes, anti-gravity all working alongside [music] each other. And that's totally different to using Hermes workspace which is just purely focused on Hermes workspace. I also saw a lot of people reporting like bugs inside our group of the Hermes workspace. That's another reason we set this up. Yeah. So the agent OS is only available for members of the [music] AI profitable room. So if you want to get my setup that you know I spend about 3 to 4 hours a day improving and coding out [music] then you can get my setup inside the agent OS of the AR proper warning. [music] So if you just go to the community here this is the only place it's available. So just go to the classroom and then [music] go to new daily updates agent OS and you can find it right here. >> [music] >> Yeah. So you can do that with there's a command now inside Hermes is called forward slasharn. So if you give forward slasharn to [music] Hermes agent and you give it the link to the guide, it can actually learn from you based on what you you want it to learn and it will add that as a skill. So, for example, if we go to Hermes here and we type forward/arn, you can see I said/arn and then I linked to one of my tutorials on the Agent OS and it said learned it and saved it for future sessions. Here's what [music] the Agent OS is, here's how it works, etc. And then it's actually saved it as a skill so that we can use what we learned from that guide in future workflows with Hermes. [music] Today I'm going to show you a powerful system that's working for us right now based on a case that I've been testing out. So you can see for example this keyword best AEO experts and we are actually ranking inside the AO views for that particular keyword. So you can see us ranking over here and if we look at the screenshot you can see it's ranking right at the top over there right and so we are ranking number one inside AI overviews for this particular keyword and then also if we look at this keyword we're also ranking on the first page here so you can see us ranking with our content here and also here so we have a video ranking and also a page on our subreddit ranking for that particular keyword. So this is a powerful system that we've used. Also, if we type in top AO experts, you can see our content is ranking over here too. And so what we're doing here is we're targeting one keyword for example like best AO experts and we're ranking inside AO views inside Reddit with videos and also articles as well. And so it's four different services, the AI's answer, and multiple different pages inside the results so that we can show up everywhere for that particular keyword. And this is pretty powerful system because if you can rank number one on Google for your target keywords, for example, like best AEO experts or for example like Hermes AI SEO, you can see it's ranking right here for those particular keywords. You can rank multiple times and then if you rank multiple times, that tends to influence the AI overview. So if we look at this particular keyword, you can see that we are ranking inside the overview here because our content is ranking multiple times on the first page for that topic. And so it's a pretty powerful system. And I'm going to show you exactly how we do it, how we do this. And the the main point I would say here is like before we were doing keyword research. It was difficult to index. It was a lot of work. We'd have a full team. And what we did instead is wired the whole pipeline into the agent operating system. And then it can just run itself in parallel. So if we go over to the SEO content pipeline here, we can do keyword research. So we can for example type in like Hermes here, then we can research on Google Workspace, see what we're getting impressions for but not clicks for. And from there we can pick a keyword and use that topic inside our content generator. So we plug in the keyword, plug in the case study, click generate articles, and we can rank across multiple different platforms. But also, as I've shown you before, we can create video content. So, if we have a look inside the video agent, here's an example we created previously. And this is a full avatar video fully generated in one single click using a system where it does the research, the script, the B-roll, and the avatar and the voiceover all inside one system. So, the way that we do this is basically we type in our prompt for the keyword that we want to rank for, create the video, and then we can start ranking. You might also say, okay, does that work in reality? So, this is a channel that we've actually built up using a similar process. And you can see it's grown pretty nicely using this full system here. And also, one thing to note, you know, some people say this might not be the best quality content, blah blah blah, or AI can't create good SEO content. I would say with the system that we've got set up here, it's a 13step system. is so unique in the way that it organizes AI to make sure it follows all the rules and be super specific that when we're creating content with this whole system here, we make sure it's as quality controlled as it can be. And this would literally take like a human a whole day to quality control each blog post cuz it's so specific. But because you can give it to Claude or you can give it to Hermes and then it creates the content step by step. It's fully optimized in a timeefficient way without you having to quality control yourself. And that's pretty powerful. system. So, it's pretty awesome. And also, if we have a look at some of these pages that we're ranking, so for example, if we look at this keyword, best AR community will rank in with our page here. This was fully created with our AI agents as well. Now you can also build full websites that actually rank like this inside the agent OS2. So for example, if we go to free clawed code and we see what we've built recently, we actually created a full website similar to this with a full blog setup here and then we can easily plug that into our SEO tool. So whatever you want to build and whatever you want to automate and rank, you can do it using this system. And the cool thing about this is it's pretty nicely designed and we've done it for free because we've actually got Omniroot which is a free API system. Plug it into free claw code and then it can create content for us. It can create websites for us. It can uh generate the blog post for us and everything else. And so the old way of doing SEO is that you would guess keywords, write one article, one site, publish, hope that Google finds it, and then you know if you were trying to rank on multiple different platforms or you're trying to create videos as well, that would be very time and energy consuming. With this system, we get real Google search console data from the research tab. That's actually using a Google Workspace um API. And then Claude can create the content and we plug in a keyword plus very important we add a case study here. So it's information gain. It's actually useful new information that you can't find anywhere else. And then from there we can deploy the content. Now if we have a look for example if we look at this particular keyword right so we're ranking on the first page for this keyword here with this blog post. If we look at this particular keyword, what we can actually see as well is that it's very very unique to me. So the content itself is talking about a case study that I've worked on. It's all personalized to me. It's all unique. It has information about me. It has unique CTAs to me as well. And so you won't find that content anywhere else. It's super unique. How are we doing that? Well, basically, we're plugging in case studies each time we create the content. And then we also have a memory system that's plugged in to make it personalized to us as well. So with those two systems, we can add information gain when we're doing SEO, which means that the content we create is much more likely to rank because we're creating something new and unique. And then additionally, we can find brand new keywords that are totally unique as well because when we're running the data for keyword research, we are using Google search console as well. We also got something else built in for rank tracking, which is open SEO. So we've built in open SEO into this whole system. so that we can check our rankings. We can do AI rank tracking and everything else and then we can deploy content whenever we need to too. And so the great thing about this is as well that when we're publishing that content, it also gets indexed automatically with something called index sectional. And that means that the content we create tends to get indexed very quickly and we're much more likely to rank. So if we have a look, for example, today is the 6th of July and this content was created on the 3rd of uh June and it's indexed. It's already ranking for that particular keyword. Now, if you're wondering how does it work step by step, so you research using Google search console data. is going to write five unique articles, deploy them, index them with indexal API, and then that's going to loop around so that we get more keyword data so that we can create more content around those unique keywords that nobody else knows about. We're finding hidden keywords that nobody else knows about because we are using the data that nobody else has access to. Here's another example. So, we actually created this blog post literally like about a week and a half ago. And you can see here that it's already getting clicks and traffic from this system. So, if we look at Google Search Console data, you can see how many impressions, how many clicks we're getting, our average position for that keyword. And the content is fully formatted, looks beautiful, includes unique case studies, adds nice tables, and everything else. And also the good thing about this is it's got internal links and external links as well plugged in as well. So it's a really powerful system. So the way it works is you just give it one keyword, one unique case study, it can create five unique articles and then it auto deploys those to our website. So like I think we're at the point now where with SEO, you shouldn't be logging into the website. You should be getting your AI agents to do that for you. Otherwise, you're going to be wasting a lot of time. And also, this means that you don't need to have a virtual assistant post it for you, which makes the whole workflow more efficient and saves you time as well. So, it's a really powerful system just using Claude and Claude AI SEO. What we also have set up inside here as well is Claude Ultra Code. So, if you want to have like teams of sub agents working together, let me show you an example of this right here. So, we actually created a task where we said, you know, build out a full keyword strategy and a full SEO strategy for our website. And you can see all of these sub agents were spawned with the ultra code system inside Claude. Ultra code is just a way of creating a task and having a team of agents and sub agents working together on that task and it just works in the background and it can work for like quite a long time autonomously using this whole process. So that's another way that you can do SEO and automate it too. And then you rank with multiple different pieces of content. AI is more likely to site you and therefore you build more authority faster as well. So if you want to get the full system and the machine that we've set up for SEO, you can get that inside the AI profit war room inside the agent OS and everything that we've shown you today. The SEO pipeline, the Google search console keyword loop, the video agent, the swarms, plus the full system that you can see right here, including the memory system as well. You can get that all inside the AR profitable looming link in the comments description or just go to the arr.com. And if you want the agent OS, just go to the classroom and you can grab the agent OS here with the last update date, the video tutorial, the zip file to install it. And we add new daily guides as you can see right here. Inside the classroom, you can also get our AI SEO automation systems here. So these are all the best systems that I've created for AI SEO. Inside the community, you can ask questions and get help and support whenever you need to. And you might also think this stuff is technical, right? Difficult to use AI agents, but actually we've got 196 pages of wins of testimonials of people setting up these systems and getting amazing results. I mean, for example, Jacob, he just said uh just set up Julian's agent OS in 10 minutes. And it's a strange feeling because he spent 12 hours a day last week trying to build the same thing. And so he's fully set up a system and saves a lot of time along the way. You can also see, for example, David is saying like Julian is cooking harder on his agent OS than anybody I've ever seen. Paul actually set up this agent OS across multiple different devices. So, they all linked together. And you can see here, Shin says, "Out of all the school channels have joined, this one by far has the most value." So, people are just absolutely loving the ancient OS. And I think it's great to have a community where you can share what you build and put it all in one place. So, if you want to get that, it's all inside the app profitable. And then also, if you like the idea of this stuff, but you don't have the time to set up, you can actually book in a one-to-one SEO strategy session, link in the comments description, or just go to goldie. Agency. And on that call, our team of experts are going to look at your website, analyze it, show you how to rank, answer any questions you have, and show you how you can actually build out a website that ranks based on what's working for us. We test this stuff and then what we do is we show you exactly how to rank for this sort of stuff using our free onetoone SEO strategy session. So feel free to book then link in the comments description. Thanks for watching. Big DAP says, "Late to the party. What did I miss?" So, today we have already covered a Q&A session. We talked about RTK and how to reduce tokens. We also ran through how to use codeex for free with Omniroot and how to use Claude code for free with Omniroot. And then also how I'm using the agent OS for AI SEO as well. And then we're [snorts] just going to run through the final um setup. Now, Today I'm going to show you everything that I built recently with Hermes agent just to inspire you and to show you what's possible and also to show you how powerful this is when you really understand how to get the most out of it. Now, one of the first things that you'll notice with Hermes is that if you're using it inside the terminal or the desktop app, it's quite limited in terms of what you can do and how you can build with it. With this system that I'm going to show you right here, it becomes 10 times more powerful. So, let me show you an example of this. What we have here is Hermes Astros. Now, this is a custom workflow I've built with Hermes where basically we've given it access to an API that can monitor our competitors, monitor certain keywords, and then break down all of the latest topics that we could create content as round around as well. Now, if we open up each one of these, for example, like this one, you can see here that it breaks down the topic, uh who created it, and then also links to the original. And it also explains like, okay, here's some titles that you could use around a similar sort of topic that unique. It gives me angles in terms of what we could talk about. And then what we can do from here is we can actually click through to our video agent to notebook or to SEO content. And what that means essentially is in one single click, we can create a unique piece of content based on what's trending on this topic. And we can plug that into our video agent to create a video about it. We could create SEO content about it directly. We could plug that into Notebook and create like a a PDF research or for example a PowerPoint presentation or a video using the power of notebook. And so this whole system is just operating 24/7, keeping an eye on what's trending right now and then helping me stay updated with all the new topics that I could create content around inside one single system. And that's just one tiny little part of this. What we also have over here is an outreach tool. And basically, this looks and feels like a SAS tool, but it's still Hermes agent. So, what we have here is a lead generation tool where I can see how many leads we've got, how many were validated, how many emails were sent, etc., who's on the list, and then also we can type into here whatever leads we want to find. So, we can generate leads in like one single prompt. Once it what it's going to do from there is pull in the latest leads in terms of who they are, how to contact them, and enrich them. So, the latest notes on them. We can delete any of them over here as well. And it also hides the big companies that wouldn't respond to our outreach emails anyway. And then from here it we can manage our campaigns. So we can automatically with AI tell the AI exactly what we're going to reach out to them about. And then we can create a campaign automatically. So we have all of our campaigns over here. If we want to delete one, we can just click delete over here or we can pause it. And that is super easy to set up. So we've got Hermes Asteros that can monitor our competitors 24/7. We have the outreach tool that can do lead generation on the spot and send emails and automatically build out the whole campaign for us. From here, we also have Hermes Apollo. So, Hermes Apollo is a voice activated version of Hermes where we can talk in real time, super fast to respond. We can see everything that we've talked about previously. And then we've also got all of our builds that we created over here. So, these are all different things we've actually built using uh Hermes Apollo. And so it can build out mini apps. It can build out stuff on the go. It can actually create quite nice websites as you can see right here. And it can do all of this in one single prompt. So if we switch this on, we can actually talk to Hermes Polo over here. And I'm just going to mute it cuz I don't think the sound will sound very good on the reply. But you can see that it's taking notes here. And if we say for example, open up Obsidian, it just opens it up, right? So it opens up literally in like 2 seconds. And from here, it can use computer use. If we go back and we say, okay, open up Google, it can use browser use as well. And it can run really quickly with that, too. If we say for example build out a beautiful to-do list app, it will start building as well directly here. Pretty powerful stuff. Now we can also switch between auto and agent mode. Agent mode is a bit slower to reply because it uses more tools, but that means it can actually do more in the background. We can also have a wake word for this as well and it can just sit in wall mode. So it can just sit in the background and then we can command it with a wake word whenever we need it. Plus, it can do like daily briefings here. So, we can switch between daily or weekly briefings and we can see everything that was talked about previously. So, what it does with these briefings is it gives us the suggested focus. It reads out the briefings. So, it's actually reading it out right now and then it talks about the open action items here in terms of what we need to do and then what we're working on and the main headlines inside our industry. So, it gives like a full weekly breakdown. Pretty amazing stuff. So we have Hermes Apollo, Hermes Astros, we have the outreach tour over here and we also have Hermes Oracle. Now Hermes Oracle basically pulls in the latest news in our industry so that we don't need to stay updated with it. And what we can do is click on any of these and it takes us directly to the news update and it breaks down exactly what it's talking about. So if we look at this particular update, we can see it was last updated 1 hour ago. It focuses on FN that's trending. If we need to refresh it, we can click consult the oracle here. And the amazing thing about this is it's super useful because let's say for example this is trending right now. Well, when it's trending, so for example, if you look at this Gemini 3.5 Pro leaks 2 million token context window. So we can actually publish an article about that in one single click. So I can create content in one single click. It will publish an article about that directly to my website. As you see here, it says writing forjulian.com. We can also draft social media content about that topic as well. And it actually categorizes the topic that's talking about. So for example, these three news updates are all about models. Now you see how it says 88, 92, 95, etc. These are basically organized in terms of what's trending the most. What is the most popular news right now? Because obviously we want to focus on the most popular news so that we can create content around that and that gets the most attention. And then if we scroll down, we get six new pieces of content per day to talk about. And if we need more, we can consult the oracle. Now, if you want to see, okay, what have we actually built previously, you can see our previous content creations over here. So, this was generated one day ago. And you can see the full article over here. It's neatly formatted. It reads nicely. It's written from first person perspective. And also, it has like nice tables and formatting here. I just generated it in one single click. also has CTAs on the page to our leads uh to our funnels and everything else. And that is all ready to go. Pretty crazy stuff. So that is Hermes Oracle. We have Apollo, we have Hermes Asteros, we have the outreach tool and we also have a studio here where we can generate videos and images and voice notes on the spot which is pretty cool. And additionally we have mixture of agents here. Now if you're wondering okay what is mixture of agents? So mixture of agents is basically a powerful way to get kind of like frontier level outputs with cheaper models. So what you can do is you can combine two different models together and then you can ask a panel and generate more outputs and better outputs here. So, for example, this game was created using the mixture of agent setup. And when we've tested it on Goldie Bench, bear in mind that Fable 5 is not going to be available on subscription after the 7th of July. You can see here that Hermes Mix of Agents actually outperforms on our test called Fable 5. So, we've run 47 different tasks. And if you're like, "That doesn't sound quite right." Well, then, you know, feel free to check out the website and check out what we've created cuz we add everything here. But from what we tested, we actually found that Hermes mixture of agents can create better outputs than Fable 5 itself. And it kind of makes sense because when you're using mixture of agents, you got two models working together and they can basically quality control each other and you get the best of both worlds, right? Whereas, for example, if you are building out with just Fable 5 alone, well, there's no one to quality control the work. there's no one to check it. It doesn't have two minds working together. And so I think a team of AR models is always going to outperform a singular model if that makes sense. So that's the way that we look at it. And if you want to see for yourself everything we've created, you can get that right here. But that was all done using the mixture of agents model that we've got plugged into the Hermes setup here. And the thing that I would say here is like if you're using Hermes inside the terminal or inside Hermes desktop, you can't customize it in the same way. So if we run Hermes here inside the terminal, you're just not going to have the same impact. You're not going to get the same benefits because if you use Hermes like this, where's your conversation history? Where's all your different workflows? Where's the outreach tool? You can't view it in the same way. It's not customizable like that. Whereas when you got this system, you can get the best of everything that Hermes does and it isn't absolutely unbelievable tool. Like really, really powerful. probably the best AI agent in the world, but only if you know how to use it in the right way and only if you build in the right systems. So this is what we're doing with this. Now on top of that, we have orchestration systems here. So we have this working with all of our AI agents. So for example, Paperclip, the AI agent group chat, we've got the pipeline where we can go from idea to implementation really quickly. And we have this all built in next to like an SEO agent, a video agent, a music agent, a game studio, image generator. We're not bookm as well on the MCP. We have a camb board for our Hermes agents so they can work together and then delegate tasks and work together and complete them together. We have for example the memory system here where we can plug in context into all of our AI agents and Hermes keeps it automatically updated to us. So for example when I was using Hermes a minute ago guess what that's just been updated inside the system here. So we've got all the conversation from Hermes. Pretty amazing. And then you can add cool stuff like for example we got GLM 5.2 two plugged into um claude code here. We have for example free claude code running with omniroot so that we can code as much as we want. We have omniroot over here which means that we can code with 90 free providers whenever we want and it's all inside the AI profit boardroom. So if you want to get this full agent OS as you've seen here we keep improving it. I spend about 3 to four hours a day improving it every single day and building new cool stuff into it. You can get that inside the air profile boardroom link in the comments description or go to the air profile.com and inside the classroom if you want to get the Hermes agent OS that we built. It's right here. We update it daily. You get a video tutorial, a full guide on how to use it and a zip file to install it. And then as you can see, we add new daily guides on how to set this up and get the most out of it. Now, inside the community, you can post questions, get help in real time, and I personally answer them with a video tutorial every day. And then inside the calendar, you can actually jump on weekly coaching calls. So we have four weekly coaching calls where you can share your screen, ask questions, meet other community members, and inside the map, you can actually meet people locally near you who are building with AI agents like you. So feel free to get there. Link in the comments description or go to the aiprofitborn.com. Thanks for watching. Let me just double check for you here. We've actually got a new version that I need to add regarding Hermes Astro. So, let me just double check on my latest version. So, this is the new version that we just created today. And if you go inside there, let's have a look. Yeah, it's inside the radar section there. Just let me double check for you though. We'll get that added for you right now. I'll do it whilst I'm here. Thanks so much. I appreciate I'm the real McCoy, my friend. Uh, which one? This one right now. That's dcript. Yeah. So, it's really good for editing. So, you can edit things just like you would with a a blog post essentially. Paul, you absolute legend, mate. Thank you very much for sharing that. That is super kind. And um well, it's great to see that you're enjoying it and you get the most out of it. So, thank you so much, Paul. Keep sharing your journey with us. All right, I'm going to get this finished off in the background. That'll be updated later inside the air profit boarding for the new agent OS. I appreciate everyone who's been on today. Uh, you know, there's so many great people on here. Jo, Paul, everyone else who commented. I've answered every single question, every single comment. So, appreciate you all and I'll see you on the next one. Thank you.