AI News Today
← All episodes
Episode 5 · May 27, 2026 · 15:36

Hermes Agent Operating System is INSANE 🤯

Build a Hermes Agent Operating System (Memory Layer, VPS Setup, Workflows & Context Tips)

The video explains how Hermes Agent has become much more powerful and shows how to build an “agent operating system” that organizes multiple AI agents and workflows in one mission-control setup that can run tasks automatically 24/7 without needing a powerful or technical computer. Using community questions, it covers creating and improving workflows (including a video workflow), adding a memory layer via an Obsidian knowledge graph so agents can store, update, and organize personal notes for more relevant outputs, and why the system improves continuously over time. It compares Paperclip to a custom Hermes Kanban-based team of agents, demonstrates running an agent OS on a VPS using Hostinger with Cloudflare Tunnel/Access for anywhere access, discusses organizing files and skills in Hermes, handling context via /new and /compact, and suggests focusing on Hermes (and Claude) while keeping things simple to avoid overwhelm.

00:00 Hermes Gets Supercharged
00:42 What Is An Agent OS
01:32 Build Workflows Fast
01:52 Add A Memory Layer
03:10 Paperclip Vs Hermes
04:07 Run It On A VPS
05:24 Organize Hermes Workspace
06:51 Delegate Big Projects
07:38 Sync Telegram And Obsidian
08:19 Context And Best Tools
09:31 Groq Talk Mode Demo
10:50 VPS Vs Local Options
12:07 Obsidian Migration Advice
13:16 Don’t Get Overwhelmed
13:59 Community Wrap Up And CTA

Full transcript

Hermes agent just became 10 times more powerful. And there's one thing I'm going to show you today that makes the biggest difference right now. People are running full agent operating systems, every AI agent organized in one place, handling tasks automatically running 24 seven from any device in the world. And the crazy part is you don't need a powerful computer.

You don't need to be technical. I've seen people set this up on nothing more than a basic laptop. And today I'm going to show you exactly how it works and how to build this. So that I'm answering all the questions I keep getting about how to actually implement this so that by the end of this video, you'll know exactly how to build your own agent operating system.

So make sure you stick me to the end because the tip I say for last is the one that ties everything together. Let's get into it today. I'm going to be running through the latest questions. I've been getting loads of questions about Hermes and everything else and agent operating systems and of course, sorts of stuff like that.

So what I actually do is answer the questions inside the community on tutorials like this. I can help as many people as I can because I know if they have these questions, you probably have these questions and so you're probably wondering, okay, like how do I implement this stuff? And for example, Ali posted they actually got his first agent operating system set up right now. If you don't know what an agent operating system is, this is basically an example.

So you have all your agents set up in one place, all organized. You can, for example, have common workflows like you can see over here. Basically the whole point of this is that you create everything in an organized mission control so that you can control your AI agents, sync them together, get them working together. Right.

And so Ali actually set up his, which is pretty amazing. The other cool thing about this is if you've never used this before is like you can, if you have a new idea for a workflow, you can set it up, right? For example, we had this idea for a video agent this morning and we set it up and created this workflow right here. And then you can see some of the outputs over here, right?

Like it's pretty amazing for doing this sort of stuff. And you can see an example of what we've created using this process. And so one of the other things that Ali recommended, and this is a great tip for anyone watching is like, if you're using, for example, Hermes agent always have a memory layer, right? How do you set up a memory layer?

You want to organize all of the notes about you and yourself and how you, it works, et cetera, inside a beautiful knowledge graph like this, right? How do we do that? We use something like Obsidian, right? Obsidian is a great way for your AI agents to connect locally to a memory, take notes on your day, export that to Obsidian.

And then the cool thing about that is number one, your AI agents have a lot of memory about you, which means you'll get more relevant outputs. For example, if I ask Hermes about keyword research, it will give me relevant keywords to me. Right. But number two, when you're doing this, your agents can update and organize it for you, right?

You shouldn't be organizing this yourself. Your agent should be doing it for you. That's the whole point. And that's what saves you a lot of time.

So great tip and well done to Ali. Absolutely awesome stuff. The other cool thing I would say about all of this is when you're doing this, you will find new improvements and you're like, Oh, I want to set that up. And then it continuously improves every single day.

The other thing that would say is I regret not setting this up sooner. I'm like, why didn't I set this up in January? Why didn't I set this up in February, et cetera. Right.

So the sooner you set this up, the better because it's going to save you a lot of time and keep everything way more organized and way more fun. Right. And then we've got questions here. Like for example, can an agent operating system operate like Paperclip?

Yeah, for sure. The thing that I actually found with Paperclip, this is why I don't use it so much anymore is like it was quite buggy when it linked to Hermes agent and Hermes agent is probably the best AI agent in the world right now. Right. And so if I can't link it properly, if it's quite buggy, I might as well create my own.

And so that's what I did here. But yeah, for sure like the main benefit of Paperclip is that you have a whole team of agents working together. And so that's what we have inside the Kanban board with Hermes agent. So I can drop a task in here.

My whole team goes off and completes it and we've got it all ready to go. Right. And then also I can plug in more agents as we go along as things evolve, as you can see. And also the difference between this and something like Paperclip is that I can plug in custom workflows, right?

So whatever I'm working on day to day, I can add a beautiful workflow into that. It takes me a few minutes. Thanks everyone for watching. All right.

If you have any more questions as we go along, by the way, feel free to ask. So we've got another question here, which is is there a way to implement an awesome agent operating system for Hermes running on a VPS on a virtual private server? So you can for sure. We've actually got an example of that from Dr.

John Meza, right? So Dr. John, he did exactly the same. So he actually built the whole agent operating system on a VPS, right?

And he deployed it on something called Hostinger. So if you're wondering about this, yes, you can. The way that this can work is you can use Hostinger with a VPS with Cloudflare, right? You can use Cloudflare tunnel, Cloudflare access, and then you can access your agent operating system from any device, right?

Which is pretty cool as well. And so here's an example of this loaded. He can access it from anywhere and yeah, that's a really powerful way of using this. Again, if you have any questions, as you go along, feel free to ask and I'm happy to help you.

And then what we do each time I answer one of these questions, I link to the tutorial and then I give you the too long, didn't read section here, right? So that way we answer your questions. Eamon only asked this 14 hours ago, we've already answered him. And then also the cool thing about this as well is the whole community's helping each other as well, right?

And for example, John solved this problem that Eamon had and we will help each other and we learn and we grow together, which I absolutely love as well. Next up, we've got another question here, which is from Omar, right? And so Omar was asking like, I have a question about the file section in Hermes workspace. Is it meant for storing files that you're currently working on or that sort of thing?

So if this is for Hermes workspace itself, what I like to do, and this is the way that I like to organize it is like anything that I've created previously. So the way that I like to organize is anything I've created previously with Hermes agent, I will store inside the workspace. And the great thing about that is that it's very easy to come back to anything that I've built, because for example, otherwise what's going to happen is you create cool stuff. For example, it could be a video or podcast.

It could be an image, it could be an app, it could be a website, and then you forget it's there. And so what you want is a way to organize your workspace so that you have your apps, your videos, your images grouped into different buckets. So it's very easy to organize and come back to things later. Otherwise you will lose track and it'd be very difficult to track everything.

And then if you want to see your skills, what I do is I have that inside the control room here, right? So I can look at my skills, I can look at the status of everything, and it's just much easier to organize. And then also Omar was asking, like if you wanted to continue using stuff that you've created locally, can you use the same chat? Yeah.

You don't need to create a new chat session, right? You can just organize it in one place. Let's see what questions we've got now as well. So Noel says, hello, I'm new here.

Yeah. Welcome here. And then we've got another question, which is actually the information. So your workflows can be set up to run a set number of tasks.

Yeah. I don't know what you want clarification on there, but yeah. So if like whatever you want to set up or implement, you can, right? So if you want to get Hermes to work on something, you've got it over here, right?

If you want Gemini to work on something, you can. And then if you have a big task, like a big project, what you do is you plug in a task over inside the Kanban board, right? And from there, your orchestration agent will triage it and turn it into subtasks that get completed later, right? So we might say create a website and then we'll break it down into like design UI and come up with the keyword strategy and start writing the blog content, et cetera.

Right. And so that's how you can delegate big tasks into small ones automatically and get them done by your AI agents. Yeah. So if you wanted to set this up, what you can do is you can, if you want, for example, like Telegram to sync with Obsidian, the memory, then what you do is you've got two options, right?

So if you're running Hermes locally, then you can just ask Hermes locally to sync to Obsidian and use Obsidian as a memory, right? And then that'll work inside Telegram as well. So it can use the information from Obsidian and work inside Telegram. The other option that you have, if you don't want to do that or if you're storing everything in the cloud is that you can use the Obsidian MCP and you link the Obsidian MCP to Hermes agent, and then it'll work on Telegram as well.

Claude says, appreciate love your videos. Thank you very much, sir. Good to see you. How do you handle context?

So for that, what you can do is start a new session, right? So if you do forward slash new or forward slash compact, that's how you can handle that inside the chat, right? So you can do forward slash compact or forward slash new, and then that will avoid the context bloat. And then we've got a question here about what's the best sort of agent or tool to use.

For me personally, I think if I could only use one to get stuff done, it's going to be Claude, right? Claude is absolutely amazing. Very smooth to use, less buggy. If you want a free version, then just go with Hermes agent, right?

Those are my top two, Claude and Hermes right now. So I would stick to those. I do think you don't want to overwhelm yourself. So stick to one thing at a time, keep it simple, stay focused, and then you can build stuff in as you feel more comfortable.

And then you can also see here, so Dr. John Meza, he actually set up Paperclip agent as well. So that's pretty cool as well. And what he did here is set up Paperclip and an AI integration engineer agent, right?

And then got the instructions from that, which is pretty cool, right? So there's some people building like amazing stuff inside this community. Yeah, pretty amazing stuff. So well done, John.

That's actually amazing. Super smart as well. Richard was asking does Supergrok light work on Hermes with OAuth? As far as I'm aware, I think it does.

So as far as I'm aware, I think it does because I'm not on the highest level and it seems to be fine, but just check that it might have very small limits. So you need to be careful with that too. But I will say either way, like from what I've tested, Hermes with Grok is probably my favorite setup right now because it can generate images, it can generate videos, text to speech, it can generate podcasts. And also you can speak to it in live mode as well.

So for example, we go to Hermes here. In fact, I actually did this with OpenCore yesterday and Grok. You can see that we have this talk mode over here, right? So we can actually speak to our AI agents, have a live conversation with them.

We can change the voice inside that as well. And then we've also got all of our history here. So if I want to see my previous history and previous things I've created, et cetera, we can view it over there as well. So it's a really cool setup.

You can see, for example, these are different things I've created previously. If you go to the voice section, we can see all these different voice notes. If you go to videos here, we can preview the videos that we've created as well. So it's such a cool system.

Once you figure this out, the other option as well that you have, if you don't want to get a subscription with Grok is that you could use the API key and the API key will probably be quite cheap if you're not using it too much as well. So that's another option. We've got another question here from Lewis, right? And Lewis says, hi, everyone.

Question on setting up a VPS versus being local. I don't have a decent or powerful enough PC. And what would the recommendation be as I would like to still use my laptop to work? How would I set up correctly?

So you've got a bunch of different options right here, but lightweight local dashboard, for example, with Hermes on a VPS, that's what John Meza did as we covered earlier. So that's one option. The other option is that you do local, but you'd probably want a better machine. So for example, like a decent laptop, but again, you can just go with the lightweight setup for now.

And then the final setup is that you could have, for example, a tail scale running, right? And then if you've got a good sort of machine that you don't use, then you can install tail scale, run Hermes on it and access it from any laptop that you're on, right? So three different options there. If it sounds to me, like you'd probably prefer Path A with just the laptop setup and the lightweight version of Hermes on a VPS, right?

But you don't need a super powerful setup. I've seen people, for example, set up AI agents like Hermes on just a Raspberry Pi, which is way more lightweight than a laptop. And bear in mind, like your laptop only needs to run a browser this way and then the VPS can do the heavy lifting for you. All right, last question.

And feel free, if you're watching this live, like feel free to ask any questions as well. So the last question is from Jason, which is should you migrate all your files to Obsidian first, right? And which agent should you focus on? So it's a good question.

Here's what I would say, right? And number one, welcome in, right? Number one, should you migrate everything to Obsidian? You don't need to, right?

I didn't when I started using Hermes, for example, you can migrate the rest as you need it, right? You could just drop like three or five reference documents into Obsidian. Maybe your top five SOPs, maybe a brand voice document, your current goals, et cetera. That's enough context for now.

And then you can build it so you can start the build up right away. You can use Hermes immediately, and then you can ask Hermes to update the Obsidian vault every day. So the vault grows as you use it and everything gets set up and then you can migrate the rest later. If you feel like you need it, right?

If you feel like, ah, it's not good enough yet, but I don't think you will feel like that. Other issue was that Jason was mentioning the open core brace consistently. I honestly think Hermes is a lot smoother. So if you're having a lot of problems, I would give Hermes a shot.

If you're like, okay, which one to focus on? And then in general, just go with Hermes in general, right? And also re-feeling a bit overwhelmed. I know you mentioned a lot of people feel overwhelmed and you feel overwhelmed with AI.

Don't over, just focus on one thing at a time, right? For example, you could just focus on setting up Hermes this week. You don't need to do everything at once. Just enjoy the journey.

Take your time. There's no rush. You're on your own timeline and just enjoy it. And that's the way that I would look at it.

So just to recap there. And then I know a lot of people worry about this, right? So a lot of people feel overwhelmed. They don't know what to use.

There's a million tools out there. Just focus on one at a time. Enjoy the journey. Take your time.

Don't rush yourself. Be patient. Don't try and do everything at once, right? The simpler you can make this, the easier it's going to be and the more enjoyable it will be.

So I'd focus on simplifying rather than trying to do everything at once. Okay. So I think we've answered every single question on there. The great thing about this is one person asks a question, everybody learns together, right?

And so we will help each other as well. So you can see everyone else is posting on there and helping each other too, which is great as well. And that's awesome. And then this is only eight hours ago that Jason posted and you can see like all the advice and the help he's gotten.

I've already created a video tutorial with him. And also the cool thing is we give you the timestamp to see the video tutorial, but then also if you post, we give you the bullet point breakdown here. So that if you don't have time to watch a video, no problem, right? You can just check out the advice right here.

So if you want me to answer your questions with a video tutorial like this, feel free to post inside the AI Profit Boarding link in the comments, description, or go to the AI Profit Boarding dot com. You will get all of my best new trainings, my agent operating system, four weekly coaching calls per week. You can connect with members in your local area. We've helped so many people inside this community.

For example, we talked about Dr. John and Ali as well, who also created his own agent operating system too. But yeah, it's pretty awesome to have a great community of people. And every time we do this, everything gets better.

So you can see all the people that we've helped here. Like so many people have never even used AI before just learning and growing. And that's what it's all about, right? Just raising each other up and supporting each other and helping each other out.

Right. And enjoying the journey as well, because sometimes it can feel a bit overwhelming or a bit scary and that sort of thing dealing with new technology. So thanks so much for watching. If you have any questions like this, feel free to post them and the links in the comments and description or go to the AIprofitable.com to get access.

More episodes

Browse all episodes →