How To Setup Your Own AI Agent: Complete Hermes Guide 2026

Want your own AI assistant that works 24/7, remembers everything you tell it, and gets smarter over time? That's exactly what Hermes Agent does. And the best part — it's completely free and open-source. This guide will walk you through setting up your own Hermes Agent from scratch. No coding skills needed, just copy and paste.

What Is Hermes Agent?
Why Run Your Own AI?
What You'll Need
Step 1 – Get Your VPS Server
Step 2 – Deploy Your Server
Step 3 – Access the Terminal
Step 4 – Run the Setup Wizard
Step 5 – Get Your OpenRouter API Key
Step 6 – Choose Your AI Model
Step 7 – Connect to Telegram
Step 8 – Set Up Security
Step 9 – Launch the Gateway
Step 10 – Send Your First Message
How to Use the Main Features
Cost Breakdown
Tips for Success & FAQ

What Is Hermes Agent?

Think of Hermes Agent as your personal ChatGPT that lives on your own server — but way more powerful than a regular chatbot. Three things make it special:

It runs all the time. Unlike ChatGPT where you have to open the app, Hermes can message you first when something happens. It can do work for you on a schedule, even while you sleep.
It remembers how you work. Not just facts — it learns your preferences and builds a picture of you across every conversation.
It creates its own skills. When you ask it to do something complex, it saves the steps and reuses them later. The more you use it, the faster and smarter it gets.

Why Run Your Own AI?

You might be thinking: why not just use ChatGPT? Here's why people choose Hermes:

Privacy — Your data stays on your server, not someone else's cloud.
Automation — Schedule tasks to run automatically.
Customization — Pick any AI model you want, cheap or powerful.
Always available — Works 24/7 without you opening an app.
Cost control — You only pay for what you use.

What You'll Need

Before we start, here's what you need:

A VPS account — this guide uses Hostinger
An OpenRouter account — for AI model access
Telegram on your phone or computer
About 30 minutes of your time

For the server, you have three options: your laptop, a home server, or a VPS. This guide uses a VPS because it solves the "always on" problem for under $10 per month.

Step 1: Get Your VPS Server

We're using Hostinger because they have one-click Hermes installation. No manual setup needed.

Choose the Right Plan

Go with the KVM 2 plan. Here's what you get: 2 CPU cores, 8GB memory, 100GB storage, for around $10/month. The cheaper plans can technically run Hermes, but they'll slow down when you add web browsing skills. Hostinger themselves recommend KVM 2 for Hermes.

Pick Your Billing Period

You can choose 1, 12, or 24 months. The 24-month option gives you the best price per month. If that's too much upfront, go with 12 months — that's the minimum to use discount codes.

Select Your Server Location

Pick the location closest to you. Lower latency means faster responses. For example, if you're in the US, choose a US server.

Backup Options and Checkout

Hostinger includes weekly backups by default. You can upgrade to daily backups if you want extra protection. Complete checkout — there's a 30-day money-back guarantee if anything goes wrong.

Step 2: Deploy Your Server

After checkout, Hostinger will show you a setup form.

Set your admin credentials: Hostinger auto-fills the username as "Hermes" and generates a secure password. Click the eye icon to see the password, then copy both and save them somewhere safe. You'll need these to access your agent.

Click Deploy: Hostinger will build your server and install Hermes. This takes a few minutes. When it's done, you'll land in the Hostinger dashboard. You might see a survey — you can skip it.

Step 3: Access the Terminal

Now we need to talk to Hermes directly using a terminal — a text window where you type commands. Don't worry if you've never used one before. It looks scary, but it's just typing text instead of clicking buttons.

Open the terminal: In your Hostinger dashboard, find your Docker project name. Copy the last 4 letters (it'll look like "hermes-agent-abcd"). Click the Terminal button in the upper right. A black window opens — you're in.

Navigate to Hermes: Type this command, replace "abcd" with your 4 letters, then press Enter:

cd /docker/hermes-agent-abcd

Access the Hermes container: Type this and press Enter:

docker compose exec -it hermes-agent /bin/bash

You're now inside Hermes.

Step 4: Run the Setup Wizard

Time to configure Hermes. Type the following and press Enter:

hermes setup

The setup wizard starts. It'll ask you a few questions:

Quick or full setup? Choose quick setup — just press Enter.
Which AI provider? Select OpenRouter — use your arrow keys to highlight it, then press Enter.

Why OpenRouter? Because it gives you access to 20+ AI models with one account and one bill. You can switch models later without signing up anywhere else.

Step 5: Get Your OpenRouter API Key

The wizard now asks for your OpenRouter API key. Leave the terminal open and open a new browser tab.

Create an OpenRouter Account

Go to openrouter.ai
Click Sign Up and follow the steps to create your account
Check your email and verify your account

Add Credits

Click Buy Credits
Add your payment method
Purchase $10 in credits — plenty to get started

Create Your API Key

Click API Keys on the left
Click New Key in the upper right
Name it "Hermes Agent"
Optional: set a spending limit, like $20 per month
Click Create

You'll see a long string starting with "sk-". This is your API key. Copy it and save it now — you won't be able to see it again after you close that window.

Paste your key: Go back to your terminal and paste the key. You won't see anything appear (that's a security feature), but it's there. Press Enter. The wizard confirms your key is saved.

Step 6: Choose Your AI Model

The wizard asks which AI model to use by default. For beginners, I recommend DeepSeek-V4:

It's cheap: $0.43 per million words in, $0.87 per million words out
It's reliable for the multi-step work Hermes does
You won't get a surprise bill

Models to avoid for Hermes: GPT-5.4 mini (struggles with tool calls) and Qwen3.x reasoning models (they talk themselves out of using tools).

Use your arrow keys to select DeepSeek-V4, then press Enter. For Terminal backend, keep it as "Local" and press Enter. You can change models later with one command, so don't stress about this choice.

Step 7: Connect to Telegram

The wizard asks if you want to connect a messaging platform. Press Enter to set it up now. We're using Telegram because it's free, fast to set up, and works on every device.

Create a Telegram Bot

In the wizard, type "1" to select Telegram, then press Enter twice to confirm
Open Telegram, search for BotFather (the one with a blue checkmark)
Click Start, then type /newbot
Give your bot a name (like "MyHermesBot")
Give it a username (must be unique across all of Telegram, like "MyHermesBot2026")
BotFather replies with a long token — copy it

Go back to your terminal, paste the token, and press Enter.

Important: If you've used another AI agent before (like OpenClaw), don't reuse that bot. Create a new one. Otherwise, both agents will fight over the same bot.

Step 8: Set Up Security

The wizard asks which Telegram accounts can talk to your agent. This is important — without this, anyone who finds your bot could use up your AI credits.

Get Your Telegram User ID

In Telegram, search for UserInfoBot
Click Start
The bot replies with your details
Copy your ID number

Go back to the terminal, paste your ID, and press Enter. When the wizard asks if you want this as your home channel, press Y then Enter. Setup is complete! The wizard shows you a summary of everything it configured.

Step 9: Launch the Gateway

The gateway is what connects Telegram to your agent. Once it's running, you can close the terminal and chat with your agent from your phone.

Start the chat interface:

hermes

You'll see your model, tools, and skills listed.

Install the gateway:

hermes gateway install

Wait for it to complete.

Run the gateway:

hermes gateway run

Getting a permission error? That's normal. The agent will show you how to fix it. Copy the command it suggests (it'll look something like sudo -u hermes /opt/hermes/.venv/bin/hermes gateway run) and paste it in. Press Enter. The gateway is now live.

Step 10: Send Your First Message

Open Telegram and search for your bot's username. Click Start. You might see "Unknown command" — that's normal, Hermes doesn't use the /start command.

Now send a real message like this:

You're running on a Hostinger VPS inside Docker. Tell me what tools and skills you currently have available.

Your agent will think for a moment, then reply with a list of everything it can do. If you got a reply — congratulations, your agent is working! 🎉

Troubleshooting

If you didn't get a reply, three things usually go wrong:

API key didn't paste correctly
Your user ID isn't in the allow list
Gateway didn't start properly

Go back to your terminal and run:

hermes doctor

It'll check your setup and tell you exactly what's wrong.

How to Use the Main Features

🛠 Skills: Save Automated Workflows

Skills are saved workflows your agent can run on its own. For example, ask your agent to do some research:

Find the top 5 YouTube videos ranking right now for "email marketing tutorial for beginners." For each one, save the title, channel, and what it covers to a file. Then tell me which gaps you noticed.

The agent will search YouTube, analyze the top 5 videos, save everything to a file on your server, and report which topics competitors haven't covered. Once it's done, save it as a skill:

Save that workflow as a skill called "competitor-watch". From now on, when I give you a search term, find the top 5 YouTube videos, save the info to a file, and call out the gaps.

Next time, just type: Run competitor-watch on "SEO tips for beginners" — and your agent runs the entire workflow without you typing the long prompt again.

🧠 Memory: An Agent That Knows You

Tell your agent about yourself and how you work:

Remember that I make YouTube tutorial videos about software, mostly for complete beginners. When you analyze gaps in competitor research, frame them in terms of what would help a complete beginner specifically.

The agent confirms the preference is saved. To check what it remembers, send: What do you remember about me?

The longer you use Hermes, the more context it builds. You'll repeat yourself less and less over time.

⏰ Scheduled Jobs: Automate Your Routine

This is one of Hermes's most powerful features. For example, schedule a weekly competitor research report:

Every Monday at 9:00 AM, run my competitor-watch skill on three search terms: "email marketing tutorial for beginners", "WordPress tutorial for beginners", and "SEO tutorial for beginners". Send me a summary on Telegram with the gaps that stood out.

The agent confirms the schedule and tells you when the next run is. It might ask for your timezone — just tell it (like "US Central Time" or "Asia/Jakarta").

To see all your scheduled jobs: Show me my scheduled jobs

To test it immediately: Run that job now

Every Monday at 9 AM, your agent researches those topics, saves the findings, and sends you a summary — all while you sleep. Other ideas for scheduled jobs:

Daily morning news brief
Weekly competitor report
Product price monitoring
Weekly business summary
Recurring reminders

Cost Breakdown

Two costs to track:

1. Server (Hostinger): KVM 2 plan runs about $10/month, includes weekly backups and a 30-day money-back guarantee.

2. AI usage (OpenRouter): Depends on which model you use. With DeepSeek-V4 and roughly 100 conversations per day, expect around $5–10/month.

Total: roughly $15–20/month for unlimited AI automation — less than one ChatGPT Plus subscription, with far more control.

How to Monitor and Control Costs

To track AI usage, open OpenRouter and click Activity
For server billing, open Hostinger → your profile → Billing
In your first month, check OpenRouter every few days until you know your usage patterns

Your AI model is the biggest cost lever. For routine tasks, use a cheap model (DeepSeek-V4, Llama). For complex work that needs more capability, switch to a more powerful model (GPT-4o, Claude Sonnet). To switch models, send /model in Telegram, pick OpenRouter, then choose your model.

Tips for Success

Start small. Don't try to automate everything on day one. Start with one simple task, get it working, then add more.
Always save as a skill. Whenever you do something complex, ask your agent to save it as a skill. Your agent gets faster over time.
Give it context. The more you tell your agent about yourself and how you work, the better the results.
Check costs weekly in your first month. Once you know your patterns, you can check less often.
Experiment with models. Try different models for different tasks — a cheaper model might work just as well for your needs.
Use scheduled jobs as much as possible. This is where Hermes really shines. Anything you do regularly should be automated.

FAQ

Can I run Hermes on my laptop?
Yes, but it won't work well. Laptops go to sleep, which breaks scheduled jobs and Telegram integration. A VPS is the better option.

Is my data private?
Yes. Your conversations and files stay on your server. The only data that leaves is what you send to the AI model through OpenRouter.

Can I use a different AI provider?
Yes. Hermes supports 20+ providers including Anthropic, OpenAI, and local models.

What if I already use OpenClaw?
Hermes can import your OpenClaw setup with one command: hermes import openclaw

Can I connect to Discord instead of Telegram?
Yes. Hermes supports Telegram, Discord, Slack, WhatsApp, and more.

How do I update Hermes?
Hostinger handles updates automatically. You don't need to do anything.

What if something breaks?
Run hermes doctor in the terminal. It'll diagnose the problem and tell you how to fix it.

Next Steps

Your agent is up and running. Here's a simple roadmap for the weeks ahead:

Week 1: Send different types of requests. Create 2–3 skills. Add your preferences to memory. Monitor your costs in OpenRouter.

Week 2: Schedule your first job. Test it, watch the results, and adjust. Add more scheduled tasks.

Week 3: Review which models you're using. Adjust for cost vs. performance. Refine your existing skills.

Month 2 and beyond: Your agent now knows you well. Responses are more accurate and relevant. You're spending less time on routine work.

The longer you use Hermes, the more valuable it becomes.

Resources

This guide is based on Matt's complete Hermes Agent tutorial from the Metics Media YouTube channel. Last updated: May 2026.

Teknoding Fathurrahman is a writer and blogger at Teknoding.com. He has written many articles about technology for 5 years. Fathurrahman is driven by a desire to help others understand technology. He makes complex concepts easy to understand, making technology accessible to a wider audience. His articles are engaging narratives that help readers understand the digital world. Fathurrahman is also an active member of the tech community, sharing knowledge and insights with other enthusiasts and experts.

Teknoding EN