Agent OS: Hermes + Oracle + Jarvis + Paperclip

So today we're going to be answering some of the latest questions on our agent operated system about Hermes Oracle. We've got, for example, Hermes Jarvis, which is a voice activated version of Hermes. We can, for example, like take the latest trending news and then we can publish an SEO blog post about it to our WordPress in one single click. We can pull in the latest news right here. We even have like an outreach tool now for doing outreach with a Hermes agent. And if you don't have the leads, you can actually use the system here to generate more leads on the spot. And we're built in goal mode. So there's some really cool stuff that we've built in lately. Show you some other cool stuff like we've built in, for example, Fugu from Sakana, right, which is on benchmarks, apparently better than using Fable 5. Now test it for yourself. But what you can see here is like it can generate some pretty nice stuff like this is like real world open world stuff as you can see right here. And it's pretty simple and easy to use. So we've already built this into our systems where basically what we can do here is we can use Fugu and Sakana inside the chat and then just generate anything you want. We also have like loop engineering built in, a full video agent, a music agent, all sorts of crazy stuff. But what I want to do today is just run through some of the latest FAQs we've had inside AI Profit Boardroom. And because I know like if people are posting about this stuff, you probably have questions about this stuff as well. And so we can all learn and we can grow and we can help each other together. So I'm going to run through like some of the latest wins, what people are posting and what's been useful, what people have learned, etc. So let's have a look today. This is pretty cool. So Joe's built out a new tool with Codex and Hermes where basically it can automate based on a topic, the images and the blog post itself. So it's kind of like an SEO workflow where you can take a topic and then plug it directly to your website right here. And he's actually posted an example tutorial of how it works. The thing I love about this is like, you know, people are just sharing and learning and growing together. And, you know, there's nothing held back inside this community. It's just a way of like saying, hey, we built this cool stuff. Maybe it'd be useful for you to build it too. And it sparks a lot of good ideas between us. A lot of my best ideas for the agent operating system that we've built and share inside the AI Profit Boardroom and comes from the requests and the stuff that comes up inside the community. Let's see what we got here. So Patrice has basically built a three layer omnichannel CRM, an automation hub that powers his travel websites and newly launched mobile app. Which is pretty insane, right? Bear in mind that a lot of these people are not coders or technical people. I'm not a technical person, but you know, people are building like this crazy stuff like, you know, the full on software or apps or like multiple websites that can automatically publish SEO content for them. And then they break down like the full stack and how it works step by step. It's pretty insane. Now, this is a good question, which is from Mark. And it's like, you know, as a new AI builder, should you learn NA10, which is good for building out like multi-step workflows? Or should you focus on MCP or what's the best way to learn and build with this stuff, right? So if I had a choice, I'd go with NA10 because it's more of a platform. Whereas, for example, something like an MCP, you can actually use in your NA10 workflows. So it's kind of like a little bit like comparing apples and oranges. And I would say that NA10 is better to learn than just focusing on MCP servers. However, if you ask the question, should you focus on, for example, Hermes or Claude or NA10? I'd probably recommend at this point that unless you're an absolute beginner, unless you're starting from zero, learning something like Claude or Hermes is going to be way more powerful because it's basically a no-code version of NA10. So those are the two options I'd recommend. If you have a choice between NA10 and MCPs, go with NA10. If you had a choice between Claude, Hermes or NA10, then I'd go with Claude or Hermes because these are like no-code versions. They're easier to build automation workflows. So Jay Parsons, absolute legend in the AI Profit Boardroom, he's sharing what he's been building with his own agent operating system. And basically what a lot of members do is they get our agent operating system from the classroom, as you can see it right here. And it comes with a video tutorial. You can see when it was last updated. It was just updated today and you get the zip file to install it too. But the great thing about that is you can also customize this exactly how you want. So if you have a look at what Jay's done here, and this is just in one single day, right? And he's been working on his agent operating system for a few weeks, but today everything clicked for him. And this is quite often the case. This is, you know, you're learning AI. You have a few problems here and there. You have to learn a few things along the way, which everyone does. You know, we're all starting from scratch. But at the same time, eventually everything clicks. It's kind of like learning a language, I would say. So he's actually built his own team of AI agents with Paperclip and launched a CEO agent that autonomously defined a three-pillar business strategy, hired two specialist agents, and built a live task board. Now, if you're not sure what Paperclip is, you can see an example inside my agent OS where basically we've got a team of agents. They can work 24-7. They can build and complete tasks, as you can see right here. And we get a checklist of what they've done. And then once it's all completed, we can view it inside our gallery, as you can see right here. And so it's a powerful way of having agents being orchestrated together when you've got multiple agents. So if you've got Claude, you've got OpenClaw and Hermes, it's like, well, how do you get them working together? Well, one of the best ways to get them working together is to plug them into a team like this. We have Paperclip. Then they can complete tasks for you. And then you can see everything they've built in one place, which is pretty cool. He's also created a content producer and, yeah, connected Paperclip inside Mission Control. And also, this is pretty crazy. He's using local Claude code for everything, which we've built in. So you can see free Claude code over here. Basically, what this means is that you can use local or free APIs with your Claude code harness. So you can still get the benefits of Claude code, but you use a free API inside it. Pretty powerful stuff. So Thomas was asking about free Claude code and how to set it up or which lesson is inside the AirProfit boardroom. So this is the open source project that we use for free Claude code. And we've already built it inside the agent operating system. If you go to this section here, you can even control it with your voice. And you get a chat and a workspace as well. So you can run it with local agents too or free APIs. And if you want to get a full lesson on how to use it and how to set it up, we've got that right here too. So there's two options really. You can go straight to the GitHub and you can use free Claude code. And we have a full training on it as well. Or you can get it inside the agent OS and it's already plugged in. So it's up to you which one you prefer to use. We've got a pretty cool win here from Goulam. So Goulam was saying like, you know, he didn't want this to sound like a sort of this changed my life comment. But basically he said three weeks ago, he was just, you know, checking if this was real, checking if it was actually useful and stuff. And he said, what got me wasn't like the trainings or the courses. It was realizing this community runs on something most people don't, which is people actually show their work. And I've shared a couple of examples of that earlier today. But I think this is just really good. Like the whole point of the community is that we learn, we grow, we keep it positive. And we just keep learning on this journey. You can see Mike as well. He was saying the same thing. It's like, it's such a fantastic way to learn. And also, I think a lot of people join for the trainings, but they stay for the friends because we all connect together. We all share what we're working on. There's a lot of like regular members who post cool stuff all the time. So it's a great place to be. I think, you know, when you're learning AI, it can be super lonely because it's very hard to find people who are deep into it. If you're watching something like this, you're probably quite deep into learning AI. And so this is a great place to connect and just learn and grow together. This is pretty insane. So you can see here, Jose is now on day five of building his own operating system. And he's been using our setup, but then tweaking it and customizing it exactly how he wants. So you can see he said, I've learned how to make a fully functional agentic OS. And not only that, I'm studying about it for my college. Now I want more ideas and to learn how I can build better workflows for self-hosted N10. And you can see what is built here. So he's taken our idea on the Obsidian vault and then plugged that into his own setup. And you can see here as well, he's plugged in the notebook LM MCP setup as well, which we've got inside the agent OS. So you can see an example right here. If we go to the memory layer, we've got Obsidian ready to go. And then we have notebook LM over here too. And you can see like, we can just generate full research reports. We can generate infographics, videos, podcasts, whenever we want for free using the notebook LM MCP. So for the next setup, I would start wiring in with N810, the MCP. So if you want to use N810 inside your agent OS, you can actually use the MCP connection. And then you could have, for example, something like Claude building your N810 workflows for you inside the CLI. For making the pipeline better or adjusting the keyword, you can go straight to the agent that set out for you and just request it to change something. Also, one thing I'd say on the notebook LM section is like, you could probably make that even more visual, which would be super interesting. That'd be quite a cool, cool tweet to see. This is something I've seen as well. Like Claude, it hasn't actually, today it's not broken on me, but like every day I've seen like some sort of situation where it says the model isn't available. This isn't working. So I've seen this every day. I think the great thing about having an agent operating system is like if one model goes down or it gets taken down, as we saw with Fable 5, you can just control the system and then switch to something else. So for example, if Claude goes down, no problem, we're going to change the Hermes. If Hermes and Claude were both down, which I've never seen before, then we could switch to like GLAM 5.2. So that's the best thing about having separate CLIs inside your agent OS. So Kenneth is asking about AI automations and basically a setup where he could, for example, automate a morning brief automation. And I'll show you what we've done in two different ways in a second. So basically the idea is to create a digest. You could look at your inbox or current news or that sort of thing, but he doesn't want to overcomplicate it. So there's two things that I've built recently that I think would interest you a lot. And I've included a link down here as well to it. So one of the things that we built recently was Hermes Oracle. And basically what this can do is take like the latest trending news on Twitter, automatically refreshes every day. It's categorized by which news is trending the most and the most important. And then also it links directly to the post about it as well on Twitter, which is pretty cool. And then what we can actually do from there is we can draft a blog post to WordPress directly on our WordPress websites without logging in based on the new news updates. So we can rank it for SEO. And then also we can easily draft like a social media post as well quickly too. And the thing that I learned from that is that Twitter is probably the best for getting the latest news and information. And you can use a Grok API with that. I used Grok with Hermes. In terms of what I learned to keep it simple is, yeah, try and simplify as much as you can. So, for example, if I was building this out from scratch again, what I would do is just go to Claude and say, hey, can you schedule in a daily morning brief using this API from Grok? Search for the latest news in my industry and give it news about your industry. Search for the latest emails in my inbox and connect it to your emails. And then just go from there. There's a connector button inside your settings to connect your emails to Claude so that it can read your inbox too. The other cool thing about that is when you ask Claude to set it up, it will just guide you through the steps based on your idea. Pretty simple. So I'd recommend starting with Claude on that because it would be the simplest, easiest place to quickly build out. Kenneth was saying about how to get the value out of a community. And what he does is he actually keeps a running notepad on the classes and homework. This works really well. Actually, one thing that I learned from studying, you know, I learned five different languages. So I learned Chinese, Spanish, Thai, and Italian. And one of the things that helped me the most was using Anki flashcards. So if I really need to learn something and embed it in my memory, what I do is I create a flashcard from that with a setup where I have to guess a word. And then every day I run through that flashcard deck. And the cool thing about that is it's based on repetition, spaced repetition, which means that if I need to learn anything, I can quickly remember it with Anki flashcards. So, for example, you could take your notepad and then plug it into Anki. That's something that worked really well for me. This is an interesting question from Ryan who says, you know, with the Jarvis voice system, has anyone set it up so we can access more of their computer? So the way that we did this, and I'll pull up Hermes Jarvis now. Hey, Jarvis, could you open up juliengoldie.com? And you can see here it opens up the website, right? Could you open up Google? And you can see here it opens up Google, right? Now, you can even, for example, open up computer apps. Hey, Jarvis, could you open up Obsidian? And there you go. So it's working in real time. So we have this set up already with Hermes Jarvis based on the latest updates where it can basically operate in real time using ChatGPT and its latest API for audio. And then it can open things up, but it can't really edit things unless we switch to agent mode. So you'll see a button for auto and agent, and you can switch between them like that. You can click between them. And basically what that means is if you switch to agent mode, then it has more control. It's a bit slower to reply, but you can edit things locally. You can actually change and adjust the permissions based on what you want it to do. And I think the key here is like, don't let it do anything you're not comfortable with. So if you're unsure, don't do it, basically. And you can set clear rules and guardrails with the agent that helps you set up this whole system. So we've got a good question from Rabbit here, which is basically where he's switched from tool to tool. So for example, he started using OpenClaw. Then he tried Hermes Agent. And then he likes the idea of clawed code and Anthropic, but he doesn't want to use APIs. And then also what happened is when he was migrating from a VPS, he actually couldn't use clawed anymore. And then he started checking out Genspar. And this is very, very common. Like some people will be watching this and going, yeah, that's me. Like I've spent so much time switching between tools, testing different things. I think from what I can see, you probably just want to focus more and lock into one tool. So for example, if you like clawed, then I would just stick to clawed. If you can't use clawed anymore, then I would switch to something like Hermes and just master Hermes. The key here is like you don't need to build your own agents and you don't need to switch from tool to tool. You just need to master and focus on one. So like even, for example, inside the Agent OS, we have all these different CLIs and tools that you can use. But honestly, 99% of people are just going to use one or two, which is great. I think it's always better to simplify and focus and move in one direction because that's how you improve your skill set. Whereas if you're switching from tool to tool and model to model, then it can get a bit messy and you don't really build a skill set because you get shiny object syndrome. So personally, looking through everything that you've done, just pick one. If I had to pick one of those, I would go with Hermes myself. If you can't access clawed anymore. And then my second pick would be GenSpark. But the main point here is like focus on one thing and just build from there. This is a good question from Alex. Alex is saying, you know, Jarvis is taking a while to reply back. But as he saw a second ago, it's really fast for me, right? It can literally run in real time. And this is a recent update that we've added. So recently we added it and you'll see it inside the new updates. We added ChatGPT and its new API for real time talk. And it's way faster to respond. So I'd recommend using that instead if you can. This took the response time from like probably about 10 to 15 seconds down to like maximum one to two. And also you can switch between agent modes and auto mode. Auto mode is a lot faster. All right. And we've answered every single question inside the AI Profit Boardroom today. So you can see the power of this stuff. You can see what people are building. You see how useful the community is as well. If you want to get my setup with the agent OS Hermes Oracle, we've got Hermes Jarvis over here. We have, for example, outreach tools. You can see we've even plugged in an idea to implementation, idea generator. Because I know most people have ideas for apps and stuff they want to build. But they're like, they just don't have the time to build them. So what we set up here is like a system where you can just drop an idea in. You approve the plan and then it just gets shipped. And you can see what was actually built in the end, right? Which is pretty cool. Plus, the great thing about this is whether you're using Hermes, whether you're using Claude or like you saw before, whether you're using all of them, it doesn't matter. Because you can plug them all into an agent operating system like that. So if you want to get my full system, you can get it inside the AI Profit Boardroom. Link in the comments description or just go to the AI Profit Boardroom.com. If you go to the classroom here, go to new daily updates. You will find the latest version of the agent OS for a video tutorial. Last update date, so you know it's updated today. And also the zip file to install it fully. Now, we also have new tutorials daily based on the new cool stuff that comes out. So you can see some new stuff that's come out recently and we just add it inside the community. I answer your questions personally, as you can see today. Every single question gets answered by me with a video tutorial every day, like you can see today. And also the great thing about this is there's always people online. So if you're stuck with something, as you can see right now, there's 123 people online, which means you can get help and support whenever you need it. And people are just sharing amazing stuff, as you can see. So feel free to get that. All my best training is inside the classroom. We've got four weekly coaching calls where you can get help and support in real time. Share your screen. Ask questions. Inside the map, you can actually meet people in your local area who are building with AI agents like this. And you can also connect with me personally and direct message me too. So thanks for watching.