# Hermes Agent - Full Setup Guide (Beginner, From Scratch)

Channel: Sharbel A.
Video: https://www.youtube.com/watch?v=CcsgednSRfE
Duration: 20 min
Language: English
Words: 3265
Transcript page: https://viewrankai.com/tools/youtube-transcript/CcsgednSRfE

---

[0:00] If you've been looking for the one guide that takes you from zero to a full working AI agent, then welcome. You've come to the right place because in the next 30 minutes, we're going to be setting up a Hermes agent together from scratch. You'll have an AI agent that runs 24/7 on its own, texts your phone the second something matters, and teaches itself new skills while you sleep. And the best part is there is no code required. you can just follow along and copy paste every single prompt right

[0:33] away. And I know what you're thinking. Isn't this just chatt with extra steps? So instead of telling you it's not, I'm going to show you instead [music] because near the end of this video, you will watch it write its own automation and text me first with nobody at the keyboard. We're going to be hiring your first AI employee. I'll teach you how to give it a place to live, a brain, a phone number, a real job, and a schedule. And I'll show you how to run the whole thing for free. Let's get started real quick so you know what we're building. Hermes is a free open-source AI agent. The software

[1:13] itself costs nothing. You only pay for the brain it thinks with and where it lives. And I'll show you how to make both of those basically free as well. as the different paid options if you want to upgrade it as well. There are three things that makes Hermes agent different from a chatbot. Number one is it runs on its own. It can act on a schedule and message you first instead of sitting there waiting for you. Number two is it remembers. It remembers your preferences, your projects, the way you

[1:48] like things done across every device. And lastly, it improves. When it figures out a task, it saves it as a skill and reuses it forever. That's the difference between a chatbot and an employee. So, let's give it a place to work. The first decision you'll have to make, and it's the one a lot of people overthink, is where does it actually live? And there are two roads you can take here, and one of them is free. The free road is if you've got a spare machine lying around like an old laptop, a mini PC, a Mac Mini, anything that can stay powered on, run it there. That's what I personally

[2:30] do. Mine lives on a Mac Mini on top of my desk. It's turned on 24/7, costing me basically nothing, but it's electricity. But what can you do if you don't have a spare machine? Well, you can rent one. Those are called cloud servers and they can be cheaper than your monthly cost of coffee. And you've got options here. So here's the rundown. The easiest one to go with is anything that is a one-click host on the Hermes agent website. You can click on deploy to the cloud and then deploy to your cloud. Sign up and from there you can sign up to the Nouse portal which is exactly that. Another

[3:13] option here that I really like is Hostinger, which will install and deploy Hermes for you in one click of a button without you ever touching a terminal. They're also the sponsor of today's video, which is great news because that means you can use the promo code Charbell to get 10% off of your subscription. To install Hermes, click the link in the description of this

[3:33] video. Then, we're going to pick a plan. Typically, anything with about 8 gigs of RAM is more than enough. So, we'll go with the KVM2 plan. We'll click on choose. To use the promo code, we're going to click have a promo code. Yes, we're going to put Charel. And watch this. Boom. Just like that, we have 10% off of our subscription. Then we're going to uncheck anything that we might not need, like a readytouse AI. We're going to be picking an AI model. Next, we are going to check instant web scraping because Hostinger offers a,000 credits for web scraping through OxyLabs. This is completely free if you

[4:10] select the 12 month or more option. Then you click continue. Then once we pay, that's going to be it. We go to our H panel. You should be here at this point. I recommend you keep your username Hermes. Then for the admin password, you reveal it and you copy it and you put it somewhere in your notes on standby. Then you click deploy. It sets the whole thing up for you. And a couple of minutes later, you're sitting on your Hermes dashboard. So let me show you that. You may get this survey. You can decide to skip or complete it. And as you can see here, our Hermes instance is

[4:45] currently deploying. Once it is done, it's going to give you a green check mark along with running as a status. We will simply click open. And this will lead us straight into our Hermes dashboard. We're going to put our username and the password that we copied from earlier and click sign in. Just like that, you have made it in. Congratulations. Now, one fair warning before we go further. This agent can write files. It can run code and browse the web on its own. And a web page it reads could try to trick it. That's a real thing called prompt injection. So,

[5:23] two rules for you to carry forward here. Number one, give it its own machine when you can, not your main one with all your personal files on it. And number two, leave approval gates on, which I'll show you how to set up later in this video, so that it asks before it does anything risky. But, okay, let's install it. This part is thankfully very simple. On the free road with your own machine, you go to the Hermes site and you download it for your operating system and then you run the installer. That's pretty much it. It will take you through the installation step by step. And once

[6:02] you're done, you will land on this window, which is the Hermes desktop app. Now, I will be showcasing the rest of the setup through the lens of someone on a cloud server, but you can easily follow along by either following the flow of what I'm doing. Meaning, if your UI looks a little different, don't be afraid. Just follow the flow of what I'm saying and what I'm doing on screen. Or you can run Hermes dashboard on your terminal which will open the same exact screen I will be on. So literally you can open your terminal and write Hermes dashboard. Click enter and boom you're on the same exact screen on a one-click

[6:43] host. If you're doing it through Hermes agents website then you sign up and pick a plan. I recommend starting with the free plan and then click on create agent. Then click create instance. Once it's done, you can open the dashboard. And there you go. You've made it. Whatever way you decide to install Hermes, you will land on the dashboard.

[7:07] This is your agent's control room. Sessions is every conversation. Models is the brain. Craw is the schedule. Skills is what it taught itself. Right now, it can't actually think yet because there's no brain connected. So that's next. This is the step most people get wrong. So, let me give you the way to think about it, not just what to click. Your agent's model is its brain. It's what it uses to think. Think of it like the engine of your car. Without an engine, your car can't do anything. And you do not need the most expensive model for everything. You should think about AI models in three tiers. The first is

[7:50] top tier. And top tier is the boss. Those are the top smartest models on the market. Think GPT6 Astra or 6.1 Astra or Claude Opus 5.5. They're extremely smart and pricey. You save these for the real thinking. If your agent has hard decisions to make, coding, anything where we want to avoid being wrong or where being wrong costs us actual money, then the second is the middle tier. And the middle tier is your workhorse. It has solid reasoning for a fraction of the price. It's great for planning and for everyday jobs. It's smart and capable, just not as smart and capable as the top tier models. For the middle

[8:40] tier, think of models like GPT 5.6 Soul or even Luna or Clot Sonnet 5.5 or even Haiku. Bottom tier is the Grunt Workers. Those can be cheap or even free models. They're fast and totally fine for doing tasks like sorting, tagging, summarizing, formatting, the stuff that doesn't need a genius. So, if there's a manual task for an AI to perform, these would be perfect for it. Here's a helpful model tier list for you with all three tiers. Top tier, middle tier, bottom tier. Now, here's the trap. A

[9:18] model being free doesn't make it smart. Most of the free ones you can actually run are cheap for a reason. So, don't hand them the thinking or at least test. see how they do before, you know, just throwing hard tasks at them. I recommend giving those the grunt work and keeping a strong model for the brain. In other words, match the model to the job. So, how do you connect one? There are a few ways. So, let me show you the cheapest first. For free models, a provider like

[9:51] Open Router has models with a free tag. They're rate limited, so you get a capped number of runs a day, but they're genuinely fine for testing and light background work. Then there are local and free models. If your machine can handle it, you can run a small model right on your own hardware with something like Old Lama. Nothing leaves your computer and it costs nothing. Just know a small local model is a grunt worker, not a genius. Then we have the pay as you go models. I recommend you use open router if you want these models. You can drop in $5 and on a cheap model that lasts weeks. This is

[10:33] what I'd actually start on. And finally, here is a hack for you. If you've already or you're already paying for a chat subscription, you can actually run OpenAI's codeex subscription on that same plan. That way you don't have any new subscriptions to pay for. To connect that you go to keys inside your Hermes dashboard. You choose to connect to your codec subscription. Log into your codeex account. Copy the pairing code. Paste it in and ta you're done. I am going to connect my GPT subscription for this tutorial so that I don't pay for anything new. But if you don't have a GPT subscription, then I recommend you

[11:15] start with any of the free models or the pay as you go models. If you're picking a pay as you go model, I personally recommend you start with DeepSeek V4.1 Flash. It's a great all-around fast and cheap model. Kimmy is another great option. GLM's another great option, but I don't want to confuse you with too many model choices. DeepSeek V4.1 flash

[11:40] is honestly a great one to start with. And finally, send it a hello. If it answers, the brain is alive and we can move on to step three, giving it a phone. Now, let's make it reach you when you're nowhere near the screen. In channels, pick Telegram. Then, scan the QR code with your phone and it spins up your own private bot. Set that as your home channel. That's where it sends you your updates. And that's it. Just like that, you're on the allow list so nobody else can talk to your bot but you. Now, here's the hack I actually use, and it's a big one for staying organized. Don't

[12:21] just DM your bot inside of the one-on-one chat you have with it. Put Hermes in a Telegram group and turn on topics. Now you've got rooms for your agent, split by job. You can have a general topic where you just throw whatever at it. Quick questions, quick tasks. You can have an email topic where it drops new leads and drafts. That one's not for chatting. It's a feed. You can have an alerts topic where it only pings you when something breaks or when there's something that you want watched,

[12:54] where it pings you, hey, this happened. You can have a content topic, whatever you want really. That way you have one single agent with one unified memory but organized into rooms so nothing gets buried. This one thing makes it feel 10 times more usable. All right, it's got a home, a brain, and a phone. Let's give it a real job. This is one of my favorite jobs that I have set up on my very own Hermes. Every morning before I'm even up, Hermes sends me one message on my phone and it's, "What's your number one priority today?" And while I'm still in bed with my coffee, I just

[13:37] send it a voice note back. Something like, "Today, I want to build an email watcher that qualifies leads and sorts them into groups wired to Jev. Group A qualified gets template A. Group B qualified gets template B. Disqualified gets the disqualified template. And if it's just spam, then ignore it. And by the time I go to sit at my desk, Hermes has already started it. Sometimes it's

[14:06] even already finished it. Imagine that. Starting your workday with the most important thing in your day already finished. I mean, to set this up yourself, you can send this prompt to your Hermes agent. Every morning at 7:00 a.m., message me and ask my number one priority for the day. When I reply, start working on it right away and get as far as you safely can before I sit down. Don't spend money or send anything externally without asking me first. And that's it. It'll ask you the same question every single morning from here on out. You can also point it at your inbox to draft replies in your voice

[14:47] ready for you to approve and send. Or you can set it to watch something that matters like your web signups, a competitor, a price, and message you only when it actually changes. And by the way, if you want more high leverage jobs, you can install into your Hermes. I made a free resource you can grab from

[15:08] the link in this video's description. You'll find a bunch of prompts you can copy paste and steal for yourself. And if you want more resources like this, then subscribe to my channel and to the newsletter. And this brings us to the coolest part of setting up a Hermes agent. Let me show you. All right, watch what happens when I hand it something to watch and let it figure out how. Keep an eye on my niche for anything new worth knowing. Build yourself a skill for the check. Keep track of what you've already shown me. Run it every hour and prepare a reel for me to film about it in my

[15:47] voice and only message me when there's something genuinely new. I gave it that one file and it went and it wrote its own skill, set its own schedule, and a few hours later, my phone buzzed on its own. That's exactly what makes Hermes agent so special. A chatbot waits for you. Hermes starts the conversation and sometimes even finishes it without your input. There are two quick things you can also do that make it feel like your agent, not just any generic one. First, it learns how you like things. Just correct it once. Tell it something like, "That's too long. Keep it short. No bullet points." And tell it to remember

[16:31] that. Now, it'll become like that in every future chat you have with it, even brand new ones. You correct it once, not forever. And the more you use it, the more you will feel how personalized a Hermes agent can get. It's one of my favorite things about it and why or what makes it so hard to switch to another system. And second are skills. When it does something well that you will want

[16:58] again, tell it to save that as a skill. if it didn't already do that on its own and next time it just runs the whole process start to finish. That's how it gets better every week instead of starting from scratch. Now, let's talk money and safety because this is where people either relax or panic. Starting with money first, the server is a few bucks a month or free on your own machine. The model is the only real variable and it's tiny because Hermes cashes most of what it rereads. $5 on a cheap model genuinely lasts weeks. A guard rail I would set though, always put a spending limit on your model

[17:41] provider if you're going for a pay as you go model or a pay as you go service. So it can never surprise you. And for safety, put up approval gates on for anything that spends money. Send something to a client or delete stuff. You can do that by telling it in any of the chats. Let it do all the prep and you approve the final click. That's how you hand it real work and still get to sleep at night. Once this clicks, there's a whole other level. You can give Hermes a whole team, different bots, each with its own job talking to each other and handing off work between

[18:21] each other. I covered that in my bot mode video, so watch that one next if you want. You can also run separate profiles, a work agent and a personal one, so they don't share the same brain. But honestly, don't rush here. Get this one agent doing one real job first because that's where it clicks. You can also watch my learn 95% of Hermes agent video if you want to master more of Hermes agent next. That video is more for intermediates if you want to level up later on. So tell me in the comments what's the first job you'd hand to your agent. I want to see what everyone

[19:01] automates first. And I put my exact setup, all the prompts, the crrons, and my model picks into a free doc. It's the second link in the description so you can copy the whole setup. And if you'd like to use Hostinger's one-click install, then don't forget to click the first link in the description so that you get 10% off of your subscription. If you enjoyed this video, then leave a like. Or if you're new to my channel, then subscribe because I've got a ton

[19:29] more content like this coming your way. Oh, and would you look at that? The algorithm gods have decided you're very likely to enjoy this next video as well. So, click it and I'll see you there.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt