Hermes Agent: Full FREE Course!

Welcome to the Hermes AI agent full course. Hermes is the most powerful free AI agent on the planet right now, but only if you set it up the right way. Now, most people install it. They stare at a blank terminal window or a chat box and quit, but that's like hiring the world's best employee and never giving them a job.

So in this video, I'll show you exactly how to fix that. I'll show you the exact six part system that turns Hermes into a real AI employee. One that checks things for you, builds things for you, sends reports to you, connects to your phone. And here's the best part.

Hermes never forgets. So you can correct it once and it does the job your way forever. It literally gets smarter every single day. I'll also show you the seven mistakes that ruin this for most people.

And mistake number three is the one that almost everyone makes. So stick with me to the end and I promise your Hermes will be a hundred times more powerful no matter who you are. Let's get into it. So most people who install Hermes get stuck in the same spot.

They install it, they open it, they stare at a blank chat box thinking, okay, now what? This course actually fixes it. So we're going to go through everything, where it lives, what brain to give it, how the memory works, how it learns skills, how to schedule work and how to run a whole team of AI agents at once. And near the end, I'll show you the most advanced setup and we'll also cover goal mode, mixture of agents, connecting outside tools and running your agent across every device you own.

Now a quick fact first, because you should know what you're dealing with. Hermes is made by Noose Research and it came out in February, 2026. It's open source under the MIT license, which means that anyone can use it and it became one of the fastest growing open source projects in history. In plain English, that means a massive number of people looked at this thing and said, yes, this is the one my friend.

So why did that happen? Well, one reason is that Hermes actually grows with you. Uh, normal AI, for example, often forgets you the second you close a tab or struggles to remember you. So quite often, like you're explaining yourself, you go over and over again, you explain what you're working on and then it just kind of forgets the details.

The next day it does have some sort of memory, but it's very, very basic. So the problem with that is like you start almost from zero every single day. That's the old way. So you were the memory, you were the manager, you repeated yourself forever.

Hermes actually flipped that. So it remembers who you are, remembers how you like things done. And when you do the same task a few times, it actually turns that task into a skill it keeps forever. So the work you put in on day one keeps paying you back.

I should call this the compound employee system. So think about how a snowball works. It starts small, but every role picks up more snow and the growth stacks on itself. So Hermes works the same way with knowledge.

Every conversation adds a little, every correction adds a little, every skill adds a little and it stacks. So a normal chatbot would reset to zero every single day, but Hermes doesn't reset. That's a whole game. So the compound employee system has six parts, the home, the brain, the memory, the skills, the clock, and the team.

If you get all six right and you have something most business owners don't have yet, miss them. And Hermes is just super basic and won't help you at all. So let's start building it now. So part one, the home, where does Hermes actually live?

Hermes runs on a computer. You have three options and I'll tell you which one to pick depending on your setup. Option one is your own computer. This is what I prefer to use.

You can go to the Hermes website, you can download the desktop app and it installs itself, or you can copy one command as you can see right here and paste it in one copy, one paste. You can actually just grab the full command of GitHub on Hermes. I would recommend that you do that. It's just going to save you a bit of time.

And if you can copy and paste, you can do this. It's literally just copy and paste into your terminal. That's it. The desktop app as well is pretty simple as well.

So if you prefer, you can just get the desktop app from Hermes and that's really, really simple and easy to use. Option two is like you could use an old laptop or a spare machine you have lying around. You could wipe it, install Hermes, leave it on. And now your agent can run all day without touching your main computer.

So if you kind of want to sandbox it, that's an option too. And then another option is a VPS. That's like a small computer in the cloud. The good thing about this is that your agent stays on 24 seven, even when you shut your laptop.

Now Hermes can run in the background on your laptop as well. But the good thing with VPS is it's separate from your personal files. So you get kind of like peace of mind built in as well. And that covers like the first worry that people have before they even start, which is, you know, a lot of people think, what if this crazy AI agent goes rogue?

Well, if it starts doing crazy stuff and that's a fair question. So if you give it its own computer in the cloud, the question disappears because it can only control what you give it. You also control permissions inside Hermes so it can read files, but not delete them or draft emails, but never send them without your approval. You set the fence and it stays inside the fence.

Once it's installed, you connect it to a messaging apps, a Telegram, WhatsApp, Discord, Slack, even iMessage. This is pretty useful because you can like chat with it to your phone the same way you text a friend. So if you're walking the dog, you send a message, your agent does work back at home. If you run, for example, an e-commerce store, you could message your agent from the gym and ask, for example, like how do today's orders work?

Now let's talk about how to pick the right brain. So Hermes is a body, but the AI model is the brain and Hermes lets you plug in almost any brain you want. And this is where most people get stuck. So listen closely because picking the right brain for the right job changes everything.

There are three smart paths and the truth is you probably end up using all three eventually. So if you already use chat GPT, for example, you can connect it straight into Hermes and start right away. One note on Claude because people ask, Anthropic restricted third-party apps so a Claude subscription won't plug into Hermes the same way. And here's the one that I've been testing actually recently is Quenv 3.8 Max is a 2.4 trillion parameter model with a 1 million token context window built for agent work and it plugs straight into Hermes.

And one of the tests behind this model, it ran autonomously for 10 days going from an empty folder to a finished working build without any handholding. Now, why does that matter for you? Because Hermes runs long tasks. You want a brain that doesn't quit halfway through the race.

Option number two is that you could use a local model and this means the model runs on your own machine. Nothing leaves your computer completely private. And the best local model I've actually tested recently is LFM 2.5 2.6 B. Here's why it's special.

It was literally trained inside the Hermes harness. The team trained it on agent work on tool calls and the exact jobs you'll give it. Most local models slow your whole setup down to a crawl. This one is fast.

It can call tools and create skills and search the web all running on your machine. That means like you could switch your wifi off, sit on a plane and your agent still works. I wouldn't use it for heavy coding, but for like every day agentic tasks, it's genuinely good. To change your model, you just go to your terminal and you type in Hermes model.

Once you've installed it, grab an API key and plug it in. And the setup that I actually run is to mix them. So you can use a frontier brain like Quent 3.8 Max for the important thinking and you hand the grunt work to LFM running locally. The boring, repetitive, high volume stuff goes to the small model fast.

The hard decisions go to the big one. So like a company where the senior person does the strategy and the junior person does the busy work. The cool thing about all of this is you don't need to be technical. So if you want to plug in a model to Hermes, you can even get Claude to set this up.

So you can go into Claude and just ask it to set this up for you if you prefer. And you can literally tell Hermes like use a strong model for important work and the local model for background jobs. It configures itself. Switching models takes seconds too as well.

So another option is that you can go inside the desktop app here and then just switch the model here. So you can click between them. You see the model bit at the top. You can just switch between them, switch through the dropdowns, select whichever one you want.

So the great thing about that is if a new model drops next month, which it will, probably will tomorrow as well. You swap the brain and keep the body. The memory and every skill you've built, nothing gets lost. By the way, if you want this whole system built out for you, maybe you're watching this and thinking I want this set up properly without figuring out every piece alone.

That's exactly what we built inside the Airproft boardroom. So we've taken Hermes and wired it into our Gentic operating system, the agent operating system where Hermes, Claude, OpenClaude, they can all plug into one dashboard with a shared obsidian memory. You get the agent aware zip file, the 30 day roadmap for getting Hermes running in your business, step-by-step video tutorials and daily updates as we improve the system. Plus four coaching calls every week where you can ask questions about your specific setup live and share your screen so we can fix it together.

We've helped over 3,700 business owners in there and plenty of them have never touched an AI agent before. Link in the comments description, or go to the AIprofitboardroom.com to get access. Now let's talk about part number three. And this is the part that makes Hermes different from everything else.

So you really want to understand this. Hermes memory has layers. The first layer is a small set of files. So one file holds facts about you, your name, your business, how you like your replies.

Another file notes, the agents keep itself, right? So as a list of notes that keeps itself, for example, what's set up, what's in progress, lessons it's learned. And these files load at the start of every session. So your agent knows who it's already working for.

And here's a clever detail. These files are kept deliberately small on purpose because if everything is memory, nothing is memory. So Hermes only keeps the facts that matter. So it stays fast and sharp instead of drowning in old chat logs.

The second layer is session search. So Hermes quietly saves your past conversations in a database. It doesn't cram them into every chat, but if you ask, Hey, what did we decide about, for example, that pricing idea last week, it searches his own history and pulls the answer. And the part I love most, Hermes updates his own memory automatically.

So every few messages, it quietly asks itself, did I just learn something worth keeping? If yes, it saves it. You don't have to manage any of this. You just talk and it gets to know you.

There's also a security check built into that loop too. Before anything gets written to memory, Hermes scans it for prompt injections and hidden characters. So nothing sneaky gets planted in the agent's brain. And then there's the level above all of that.

And this is how I actually run it. So you can connect Hermes to Obsidian, a free note-taking app and turn it into a second brain. Every piece of research, every project, every output gets saved as readable notes that you can open yourself. Your agents can read those notes back later and build on them.

And here's where this gets really powerful. And it's almost something nobody talks about when it comes to Obsidian. So Obsidian isn't just locked to Hermes. For example, if you look at our agent operating system here, the vault is the shared memory for every agent in the system.

So Hermes, for example, saves a skill or a piece of research into the vault and then Claude Code can read it. OpenClaude can read it. If you switch to a completely different model tomorrow, it will read from the same vault too. So your memory stops belonging to one tool and starts belonging to you.

You can test that out yourself. You can ask Hermes to check your vault and suggest content ideas based on what's in there. It will come back with ideas pulled from your actual projects, your keywords, your actual business personalized because it knows you. If you're a freelancer, that means like every client brief, every draft, every piece of feedback lives in one place.

Every agent you use remembers. Now part four is the actual skills. If memory is what your agent knows, skills are what your agent can do. And this is the feature that made Hermes famous.

So a skill is basically a recipe, a step-by-step instruction sheet for one task. Hermes comes with skills built in managing notes, doing research, creating files, lots more. But the magic is that it builds new skills from working with you. Here's how it plays out.

So let's say for example, you ask Hermes to research your competitors. It does an okay job. You give it a feedback, check the newest posts, compare them to mine, keep it short. It does it better.

You do this a few times. Then Hermes notices the pattern and says, should I say this is a skill? You say yes. And now that entire process with all your feedback booked in and baked in is saved forever.

So if you have a look here, for example, it's saved the skill. It's written the skill inside our system. Next time, for example, if you say run by a competitor check, it does a whole thing your way first try. There's also a faster way to teach it.

And this is one of my favorite tricks in the whole system, the learn command. So you can do forward slash learn inside Hermes when you're talking to it. And when you do that, you hand it a guide. That could be a tutorial, documentation page, an article or something like that.

And Hermes reads it, works out the process, writes itself a skill file on the spot. I've actually taught it new capabilities in a couple of minutes this week. So think about that. Every guide on the internet just became a training manual for your employee.

If you found a great article about a new tool this morning, your agent knows how to use it by lunch. So if you think about what that means with like a normal chat GPT conversation, you have to correct it forever. You get the same mistakes every week, forever. With Hermes you correct it once and the correction sticks every week that passes.

Your agent knows more recipes. That's the compound part of the compound employee system doing its work because Hermes even keeps its own skill library clean. There's actually a feature called the curator that watches for skills you never use. That could be, for example, unused skills for a month.

It will get flagged unused for three months. It might get deleted. Your agent tidies his own desk. And part number five is the clock.

So this is where Hermes stops being a chat and starts becoming an employee. So a chat bot, for example, would write and wait for you to show up. So it's going to respond to your messages. If I go to chat GPT or I go to Claude, it just waits for my messages, right?

Hermes will show up with the work already done. So the feature is called cron jobs, which is just a nerdy name for scheduled tasks. You tell Hermes do this task every morning at eight and it does it forever without being asked twice. Let me show you an example.

So we have, for example, Hermes Muse here. We have Hermes Astros. We have Hermes Oracle. They're all scheduled tasks with a nice UI.

So every morning Hermes can check what matters to you and send you a summary to your phone before you're out of bed. If you're an agency owner, for example, that could be like overnight client email sorted into urgent and can wait before your first coffee. But let me give you two smarter versions because the morning brief is just the beginning. The first is exception watching.

Instead of sending you a report every day, the agent watches quietly and only the messages that come through to you are when something changed that matters. It could be like a review. It could be a competitor posted something big. If you have a look at this example, basically on a schedule, every 24 hours, you'll find the latest headlines in the AI industry.

And then it actually gives us the source of that news. It orders it in terms of priority. And this just runs without me. Plus we can check the previous runs over here.

This gets plugged into our memory system and then it's all ready to go. So here's another example. This is Hermes Muse. I've got Hermes Astros.

Now Hermes Astros can watch your competitors on a schedule and it compares this week to last week and reports only what's changed, what they published, what they launched. It looks at, for example, certain keywords in our industry. And so this is running 24 seven and I get like notifications once it's done. It's just running the background.

So it's very, very proactive because we've given it the schedule tasks and then it'll just go off and do it every single day. And we can change it. We can edit it. We can update it with new competitors and new keywords to monitor, but it's running on a schedule every 24 hours.

And that's the beautiful thing about the system. Now, also one thing to note here is like you want to clean out our jobs sometimes now and then, because if you have too many scheduled tasks and too many tasks running in the background, that would slow your agent down. So you probably want to go in like once a week, pause anything you're not using and your agent instantly gets faster. If you're wondering how to actually set up a scheduled task, you literally just talk to it and you say, set up this scheduled task and do this every day.

Right? So if I wanted to, for example, get Hermes agent to check the latest AI automation news every day, just tell it, listen, do this every single day. And then it will schedule it in. Now let's talk about the advanced level, the team, right?

This is the advanced level and it's where things get genuinely fun. So I'm going to give you a few examples here and each one builds on the last. First off is sub agents. When Hermes has a big job, it can split the work and spin up helper agents that run at the same time.

So let's say for example, you're researching a new offer for your business. One sub agent studies your competitors. Another checks what people are searching for. A third reviews what you've already made.

Then the main agent can pull it all together and give you an answer. So whether, for example, like took you a full afternoon comes back whilst you made lunch. And one tip here is give each sub agent a narrow job with a clear output. Now, if you're wondering how to get or set up sub agents, you can literally just go to Hermes and you say, okay, create a team of sub agents to do this task.

And you give it narrow instructions on exactly what you want. Now this is also where a model like LFM 2.5 works really nicely because your sub agents doing the grunt work whilst you run on the small, fast local model and the main agent then thinks with the big brain. Now we've also got profiles. A profile is a separate Hermes with its own memory, its own skills, its own job.

So you can run a content agent that knows your voice, a research agent that knows your market and an operations agent that knows your systems. One assistant shouldn't hold everything in one brain. The same way one employee shouldn't run every department. Profiles turn one assistant into a small team and each profile can run a different model.

So your content profile, for example, can run on the big brain, your background research profile on the local one. And third, we have the Kanban board. So Hermes has a built in task board like Trello. You can add tasks as cards, your agents claim them and you watch cards move from to do in progress to done.

So instead of like staring at a wall or walls of text in a terminal, you're looking at a clean board, showing your AI team working through your list. So you've become the manager and the agents are the staff. Next up, we have goal mode. Now this is the one that feels like the future.

So instead of giving your agent a task, you give it a goal, a finished outcome and it keeps working step by step until the goal is actually met. And it's not like one reply and done. This is like hours of work if it takes hours. So you set the target, you walk away and this is exactly why the brain choice actually matters so much.

So if we come back to Quen 3.8, that was built for long autonomous runs. So when you pair it with goal mode, you can hand over a genuinely big piece of work in the morning and come back to it finished. One more tip as well, before you run a goal ask Hermes to help you write the best possible goal for what you want. It writes its own instructions better than you will.

Fifth, we have mixture of agents. Now mixture of agents is a feature where Hermes takes your tasks, sends it to more than one model at the same time and fuses the answers together into one better output. So two brains attacking the same problem, then merging the best of both. When the task really matters, this squeezes out quality that no single model gives you alone.

The whole team structure is exactly the model our agent OS runs on. So the Kanban dispatcher hands tasks out, goal mode keeps agents pushing until the job is actually finished and the Obsidian vault means every agent shares one memory. Hermes handles the learning, the other agents handle their specialties and the OS coordinates all of it. Now, before we get into the mistakes, three advanced extras take this from like powerful to ridiculous.

Extra one is connecting outside tools. So Hermes supports something called MCP, which is just a standard way to plug outside apps into your agent. That could be your email, your calendar, your notes, your image tools, automation platforms. Once connected, Hermes can use them the way an employee uses the software on the computer and the setups let you control exactly what it's allowed to do.

So it can find emails, it can write drafts, but not send or delete. You choose the permissions app by app, which means the trust question isn't a leap of faith. It's a settings menu. And extra two is voice.

So you can send your agent voice notes instead of typing. Maybe you're out there walking, an idea hits you. You record like 30 seconds of rambling into your phone. Your agent transcribes it, pulls out the actual ideas, tags them and files them into your vault.

We actually have something set up as a custom workflow. It's called Hermes Apollo and it's a voice agent. And it means like basically I can give it loads of ideas. I can talk to it and then it will just handle everything and do it for me.

So for example, if I'm like, okay, open up Obsidian, it opens it up instantly as you can see right here. So it's pretty cool how that works. And it works really, really nicely with the ChatGPT real time API. Now extra number three is running tail scale, right?

And this actually connects your whole agent operating system to your mobile or anything else. Now, if you just want to connect Hermes directly to your phone, for example, you can do this pretty easily inside Hermes dashboard. So if you go down to channels inside your Hermes dashboard, you'll find it right here. If you're wondering how to open up a Hermes dashboard, you can just go to your terminal and then you'd start a new terminal tab here, type in Hermes dashboard like so, and then it will open up for you and you just go to channels to add and configure whichever channel you prefer to use.

But with tail scale, basically you could run this on a totally different machine. You could run Hermes with the whole agentic operating system on your mobile, however you want to do it. Now let's talk about the seven mistakes that make Hermes actually worse because the fastest way to ruin this is to set it up in the wrong order. Mistake number one is adding too many tools before you know what you want it to do.

So start with the tools your first workflow actually needs. Mistake number two is saving everything to memory. Memory is for the facts, the manner, not a dumping ground. Mistake number three, using a weak model for hard work.

A small model failing and retrying five times wastes more than a strong model doing it once. Mistake number four, turning every idea into a scheduled task. If a job doesn't end in a decision or an action, you don't need to schedule that task. Mistake number five is trusting sub-agent output without checking it.

So you still want to check the summaries and also don't believe everything the agent tells you because AI is still at a very early stage when it comes to that. Mistake number six is building like loads of different specialist profiles before you have one working workflow. Just keep it simple. You can do workflows first and you can do profiles later.

And mistake number seven is treating Hermes like a magic employee. It's powerful, but it runs on good instructions, good memory, good skills. That takes time to learn. So feed it well and it improves.

Feed it chaos. And it compounds chaos. Now let me clear up some limiting beliefs that most people use to hold them back. Let's break them right now.

So some people will say, this sounds amazing, but I'm not technical. I don't code. This isn't for me. But if you look at what we actually did in this course, you just like copy one command, you run a dashboard, you type a few messages in plain English.

When something needed configuring, you can actually just ask Hermes to configure it itself. And it did. You can use the Hermes dashboard and it's literally just like a plug and play UI here. So if you want to configure the phone or the MCP or the skills or the scheduled tasks, it's all inside one beautiful place.

The entire tool was designed so that talking is the interface. If you can like send a text message, you have every technical skill this requires. The second one people say is like, I don't have time to learn another tool. But if you flip that around, the whole point of Hermes is it gives you time back.

So you, for example, set up one morning brief like we did with Hermes Asteros. And then you stop starting your day in your inbox, right? You teach you one skill and a weekly task disappears from your plate forever. You run one goal mode session and an afternoon of work happens without you.

So the setup takes an afternoon. The tasks it takes over stack up for months. And waiting because you're busy is like saying you're too busy to hire the person who would make you less busy. It doesn't make sense.

So here's your simple plan for the next seven days. Day one, install Hermes and connect one model. Day two, connect it to your phone through Telegram or WhatsApp, whatever you prefer and run a real task. Day number three, take a task you repeat every week and teach it as a skill or use a learn command on a guide you love.

Day number four, connect the one tool that matters most for your work. Day number five, create one scheduled task that gives you a decision, not a data dump. Day number six, try splitting a research task across sub-agents. You can just ask it, or you could try running your first goal mode session.

And then day number seven, create your first specialist profile. One useful workflow, then another, then another. That's how the system works. And if you want to skip the trial and error completely, this is exactly what we do inside the aircraft reporting.

That's what it was built for. So you get the agentic operating system with Hermes pre-wired in, the zip file, the video walkthrough and the 30 day roadmap for turning Hermes into a system that saves you time, generates leads and gets you more customers. You get daily tutorials showing you Hermes workflows step-by-step from your first scheduled task to full multi-agent Kanban setups with goal mode. You also get four weekly coaching calls where you can share your screen, bring your actual Hermes setup and get it fixed live.

Plus a prompt library full of the exact prompts we use to build skills and scheduled tasks. You also get a member map. So you can meet other people running Hermes near you with support around the clock, because someone's always online ready to help you. Also got over 3,700 business owners inside there.

And many of them went from like never touching an agent before to running the first automated workflows. Link in the comments description, or go to the AIprofitbomb.com. Here's the thing to remember from this whole course. Every day you use Hermes, it gets a little better.

Every day you don't, it stays at zero. The setup takes an afternoon and the only real gap between you and the people already winning with this is that they started. And the person who started three months ago has three months that you can't get back. So install it today, give it one job, then watch what one job turns into.

Thanks for watching.