Hermes Agent OS 3.0 Just Dropped... 🤯

So today we have a lot of questions about the agent operating system for Hermes that we've built out. And this is basically like a mission control where you can control all of your AI agents in one place. So we've got like Codex, Antigravity, we've got Claude Code over here, we have Hermes, OpenCore, Claude, et cetera. We have all of our tools built in.

It's a really powerful system, but we get a lot of questions about this, which is what I'm gonna answer today, right? And so these are all the questions we've had recently inside the AI Profit Boarding. And I'm just gonna work through them and also help you as much as we possibly can on this. So let's talk about this and have a look where we're up to.

For example, we've got one question here from Scott who says, Hermes on my personal machine, how to use it? And he says, new to the space, look forward to making connections. Silly question, how can I install Hermes on my personal machine or do I need a separate machine, et cetera? So number one, welcome Scott.

But the short answer is yes, you can run Hermes on your personal machine. Most of us do, right? The use of separate machine comes from people who get nervous early about touching the agents and the main file system, which is fair. But the way Hermes is built is like, you can scope it to a project folder and that's all it sees.

So it's not like running rampant across all your files, right? And you can have, for example, OpenCloth sandboxed as well. So that's a totally separate flow. But yeah, you don't wanna quarantine the tool, just quarantine the folder, right?

And start small. You can have one project, work inside it and expand as you trust it, essentially. Let's have a look at the next question here. So browser is saying, hey guys, I'm trying to find the full setup for the skill and setup guide for SEO, essentially.

And he said the whole agentic operating system you built is growing to a beast level. I'm trying to build the base for it. And so basically the agent operating system has really improved. And the one that we created seven days ago isn't nearly as powerful as where it is right now, right?

Now for the SEO content pipeline specifically, the full read along is over here. And for the setup, I would just give yourself a bit of time, right? Now, if you're focused specifically on AI SEO with agentic OS, we actually have a full zip file that over here, right in this section. So if you wanna get that, I'll link to it directly over here for you.

Next up, we have a question from Andreas, which is where's the productivity board from Trello, right? And so if you wanna get access to that, let me find it for you. So this is what it looks like. It's basically my productivity system that I use.

And let me just test it out here. I'm just gonna check this out and you can get that right here, right? So you can get the Trello productivity board if you want that too. Let's see what else we got here on the questions.

Oh, this is interesting. So how do you build an agent operating system in the cloud, right? So if you have an agent operating system and it's running locally, how do you share it in the cloud and that sort of thing, right? So you can use like a VPS.

You could also use a container and you could also use remote access, right? So you could actually use like Cloudflare with Tailscale and that lets you use mission control in the cloud. But just be careful with this stuff as well cause you don't wanna, you wanna make sure it's fully secure and no one else can access it. So when you set this up, it's awesome for accessing remotely.

Just be careful that you don't share it anywhere you wouldn't, all right? And don't give it access to anything that you wouldn't feel comfortable giving access to. Let's see what else we got here. This is pretty cool.

So could you build a backlink builder with Hermes, right? So if you're using Hermes agent, how can it build backlinks for SEO? And if you're not sure what backlinks are, basically they're clickable links from a website to yours. The more relevant authoritative backlinks you get, the more likely you are to rank on Google and ever else.

So there's three ways you can do this, right? And you can pick whatever helps you the most, right? So number one, you could do outreach automation. So Hermes can actually draft personalized emails from a list of prospects.

You approve, you send, and the personalization is really what you want there. Then you've also got guest post pipeline. So Hermes could actually find blogs in your niche, contact editors, draft angled pitches, and that's probably more likely to be accepted, right? And then the final step is like, you can actually, for example, have Hermes monitor opportunities on Harrow, draft responses in your voice, and then you reach out to more people daily, right?

And then of course, you don't wanna focus on anything that's outside of those three things, right? We've got another question here, which is how can you generate content for e-com websites using agent operating systems, right? And so this is a really good example of how you can use the agent operating system, but I would use agent operating systems, not NA10, right? So Adam said, but how do I generate content?

Should I use NA10, et cetera? And so the flow that I would do is like, you plug a CSV of titles and the products you have into agent operating system, whether that's Hermes or Claude. The agent operating system can then read each role. You can generate SEO optimized titles and descriptions for each product, and then you can upload that to Netlify or just get a CSV out and upload that, right?

And then you could generate this for all your products. So that's basically what I'd recommend for doing this. You could actually use the, I think there's a Shopify CLI as well. So you could connect your agent to Shopify if you want to upload directly.

Next question here, best model for Hermes. Would it be ChatGPT? Would it be Grok? Gemini Anthropics.

So shout out to Tony for asking this question. So honest answer is here, like it does change every few weeks, but here's what I'd recommend. So right now my current favorite is Grok for agentic tasks. So it's got the best speed to quality ratio of everything that I've checked out, right?

So for example, here we've got Hermes plugged into Grok. And then if we go over to studio, we can generate videos, text to speech, images, and it's pretty fast as well. And also it can search Twitter. Claude is pretty powerful, but it's expensive as an API.

ChatGPT is awesome because you can use the OAuth. So if you already got a subscription, and then if you want free, you could go with our alpha and open router, right? And that's pretty easy to set up as well. And then Tony was asking, should you use OAuth for API?

So you can use OAuth on ChatGPT and Grok. Those are the ones that you want to focus on, but you can't use Chat, you can't use Claude on OAuth with agents, just so you know. Next up, we've got a question from Asif, who says, I'm a newbie in the community, just had a question about connecting Hermes to Claude, right? Can you do that with OAuth, et cetera?

So you actually can't. If you're using OAuth with Claude, so Claude was actually stopped with using OAuth a few weeks ago. So this has changed, but you can use Grok OAuth with Hermes, and you can also use ChatGPT OAuth with Hermes. So either of those would work, and you can log in directly into Hermes and use those instead.

By the way, if you want me to answer your questions, just like this with video tutorial, and also a step-by-step breakdown of each one, then you can join the AI Profitable Boardroom, link in the comments description, or go to the AIProfitableBoardroom.com, and I answer these questions every single day, like you see right here, just to help people as much as I possibly can. And also, I know if these people have questions like this, then probably everyone watching has these questions too. So Femi says, the new anti-gravity is good, do we still need Hermes? And I love this question, right?

Anti-gravity is great, it's highly agentic, it's browser native, it's fast. Google created a really good product there. But here's the framing that I'd offer, right? Hermes is open source, right?

So you can pick the model, you pick the host, you pick the way that you use it, et cetera. And I would say anti-gravity is great for generating, for example, websites and that sort of thing, but Hermes is more agentic, right? So if you look, for example, at anti-gravity, it's better for generating coding projects. Whereas if you have a look at Hermes, it's more for actually doing stuff.

So for me personally, I prefer both, but I'll use anti-gravity for coding and building out websites and apps and tools. And then I'll use Hermes agent for actually just going off and doing tasks. So anti-gravity for coding and building stuff, Hermes is more for actually doing daily tasks on a schedule. And also Hermes is open source, which means it's more customizable.

We've got another win here from Clovis, who said he's using the Obsidian Vault on his agentic operating system. So a lot of people are actually using the agent operating system and then sharing what they've built inside their operating system too. For example, like John as well, he built out his own operating system and you can see here Clovis too. So pretty awesome system.

And basically he got it to integrate inside itself. Now what this means basically is like, you get more memory inside your agents and you get better memory. So for example, if you look at all of our agents plugged in here, they can all link to our memory system over here. And so all the knowledge, all the context that we have is linked to our Obsidian Vault.

And so if it's linked to our Obsidian Vault, then it can create better outputs for us. It understands us better, it has better context. And that's what the game is really about. If you want super smart AI agents.

We've got another question here from Ryan, who says Obsidian is a memory layer for AI agents, which is what I was just talking about here. And so you're saying like, he's exploring the idea of having a second brain with Obsidian as a memory layer. A few things that he's trying to figure out, what vault structure actually works? How do you give access to your AI agents to Obsidian?

Any sort of tagging conventions that work well? And how do you stop AI agents from giving the Obsidian Vault like low value content? So these are my answers, right? And basically, we have something called the infinite context engine that I use.

And that basically gives me the best possible outputs I can get, right? You can see over here. So this is the infinite context engine. We have a full video tutorial and setup guide on this.

But basically, here's the way that I would look at it, right? So in terms of the vault structure, when you start writing into it, you can use either Para or Zest or Kasten. Those are two good options. They're built for humans browsing manually, right?

For agent readable, you probably want a flat structure. So one folder per project, plus a tags-based discovery layer. And then agents can use tags way better than nested folders, right? So it depends if you're using it for agents, then just a flat structure is pretty good.

How to expose a vault? So you can actually just have this stored locally, or you can use the MCP server, which is Obsidian MCP. With tagging conventions, you can have basically ones for each part of your project. So if we have a look over here and we look at the tags that we've used, and let's say, for example, we have a look at this system here, right?

You can see how we tag these depending on the project. So for example, if something is related to the AI Profit Boarding, my agents will tag it as that. And then you can also have a review agent that runs nightly if you need to improve the outputs, right? So it can read new nodes, it can flag low signal ones, and that's a good system for monitoring it.

It's quite advanced, but if you really want to go in granular detail, you can do that. Let's see what else you've got here. So John is asking about like, how to basically get an AI as a voice agent to call your leads. Now, obviously your leads need to opt in if you're going to call them with AI, right?

So that's really important, but yeah, this could work. You could use something like VAPI, right? And then the trigger could be a webhook where they get a call within 30 seconds. And it's always going to help if you get in contact with your leads as well, right?

And then also you want some time of day, as well for that, right? So that's the way that I would look at it. And also I think if it's good enough, like the lead doesn't really care if it's AI, they just care that someone showed up fast. Next one from Jason, which is about how to use agent operating system on Docker as a mobile app, which is a really cool one.

So Jason was asking about to go through the setup process and stack agents. I would be careful of just giving it remote access, but if you do want to set this up, but you could use Docker or something like that, Cloudflare, Tunnel, and Tailscale might be a better option as well. And then for the mobile app, you could, I think you could do that through Tailscale actually, that could be a good option. And I think a lot of people are using Hostinger for this.

That's what I've seen. Memory and context in larger AI agent systems. So Tony was asking how are people handling memory systems and that sort of thing. So here's a way that I do.

I have a short term way. So you can have per session context, just lives in the conversation. Then you can have midterm, which is you could have Obsidian, for example. And that's the way that I would look at it.

It's like Obsidian Vault with Markdown files, which is easy for AI agents to query and humans to read. It's probably one of the best ways to do this, right? And then also, if you don't want to update it too much, every 10 to 15 turns, your agent can then summarize the conversation or you can do this daily as well. So that's a good way to do it.

And we also have a full setup inside the Infinite Context Engine system, which I'll link to you in a sec. So you can see the Infinite Context Engine here. This is the full setup that we use. So you can do this daily or just one daily summary as well from your agents to the memory system could work really well too.

Next question. So Augustine was saying agent operating system. A lot of people want to sync this with the Claude app and that sort of thing. So the way that I would probably do that is sync it to my Obsidian Vault with the conversation summaries if you don't want to use the API, right?

So if you try and sync like Claude inside the terminal to your agent operating system, it doesn't work so well. So if you want to sync it to your Obsidian Vault and you don't want to sync it inside the agent operating system you can do that instead, right? So for example, if I go to the agentic mission control here and I try and speak to Claude, that's not going to sync to my terminal and vice versa, right? But the cool thing about this is that we can just have it summarized to our Obsidian Vault and then the conversations get logged and synced between them as a context in memory.

So that's the way that I would do it. You don't really want to sync sessions. You want to sync the memory. That's the most important part.

So I talked to a guy, he's an absolute legend. And he was saying, how do you transfer chat GPT and perplexing knowledge over to Obsidian, right? How do you do that? So let's say for example, you're on chat GPT and you want to sync all of the knowledge inside chat GPT over to your agent operating system or to your Obsidian Vault, right?

How do you do that? So here's a way that I would do it, right? You can actually migrate your old AI history into Obsidian. So Hermes can read it all in one day and start from scratch.

So the way that you can do that is you can export from chat GPT directly. If we have a look at that, and you can do this with a prompt as well. So you can actually ask like chat GPT to just tell it everything you know about you. Or if you go to this section here, you can export your data from chat GPT and any other tool, right?

And so that's where you can do it. And then once it's in Obsidian, Hermes can read that vault locally or via MCP. All right, so we've answered all the questions. It's 15 different questions from the AI Profit Boarding.

This literally just lasts 24 hours. So as you can see right here, like I just, all the questions that we have, I know that everyone has them. It's not just the people who've asked. And so that's why I answer these videos in tutorials like this, just so that I can answer in granular detail, everything that we've talked about today and just get you the best information you can on this stuff, right?

And so if you want this sort of stuff, you can bring your question, get it answered. And every Q&A is built from questions members posted inside the AI Profit Boarding. So if you've got one, maybe that's Hermes setup, agent operating systems, Obsidian, SEO pipelines, anything, drop it in the AI Profit Boarding and we'll cover it in the next session, right? So this is all designed for people inside the community.

You can connect with me personally. You can ask questions inside here. You can also jump on weekly coaching calls. You get all of my best systems, including a new version of the agent operating system that we just dropped this morning with all of my new updates.

You can connect with people inside the map here as well. And everything is right there for you, right? And you've seen all the wins here as well, right? We've got so many cool testimonials and wins from people who've never done this stuff before and have just learned how to do it.

So that's what it's about. Link in the comments description or go to the AI Profit Boarding.