Udemy
    •  
    •  
    •  
    •  
    •  
    •  
    •  
    •  
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
Build a 24/7 Self-Hosted Personal AI Agent on a VPS(Hermes)
Rating: 5.0 out of 5(1 rating)
14 students

Build a 24/7 Self-Hosted Personal AI Agent on a VPS(Hermes)

Set up Hermes on a VPS, text your agent on Telegram, watch it in a dashboard, and automate tasks with cron
Created byAlpha Omondi Ogilo
Last updated 6/2026
English

What you'll learn

  • Deploy a personal AI agent that runs 24/7 on your own low-cost VPS
  • Harden a Linux server with a non-root user, key-only SSH, a firewall, fail2ban, and automatic security updates
  • Install and configure Hermes, connect an LLM provider, and run multiple agent profiles
  • Operate your agent from the CLI and TUI — sessions, tools and toolsets, memory, and context files
  • Install ready-made skills and write your own to extend what your agent can do
  • Connect a Telegram bot through the gateway so you can message your agent from your phone, anywhere
  • Monitor your agent in the Hermes web dashboard over a secure SSH tunnel
  • Connect external tools with the Model Context Protocol (MCP) — wire up GitHub and manage issues from Telegram

Course content

6 sections • 31 lectures • 2h 49m total length

  • Introduction/Demo6:37
  • Concepts3:42
  • Simple Architecture Walkthrough2:44

Requirements

  • A computer running macOS, Linux, or Windows (with WSL2) — anything from the last 5 years is fine
  • Basic command-line comfort: you can cd into a folder, run a command, and edit a config file. You don't need to be a sysadmin — we teach the Linux server side from scratch
  • A DigitalOcean account (or any VPS provider — Hetzner, Vultr, Linode all work).
  • A credit or debit card to fund the VPS and an LLM API key
  • An LLM API key from at least one of: Anthropic, OpenAI, OpenRouter, or Nous Portal. OpenRouter is the easiest first choice ($5 minimum top-up gets you through the whole course)

Description

You will set up a personal AI agent that runs 24/7 on an affordable server you fully control. One you can text from your phone and watch work in your browser. Over 30 hands-on lessons you'll go from an empty Linux box to a hardened, always-on agent powered by Hermes, meeting it two ways: message it on Telegram like a friend, and open a private web dashboard to watch its sessions, memory, and tools. By the end it runs on its own — gathering a briefing every morning and texting it to you before you're even up.

What you'll learn:

  • Deploy a personal AI agent that runs 24/7 on your own low-cost VPS

  • Harden a Linux server with a non-root user, key-only SSH, a firewall, fail2ban, and automatic security updates

  • Install and configure Hermes, connect an LLM provider, and run multiple agent profiles

  • Operate your agent from the CLI and TUI — sessions, tools and toolsets, memory, and context files

  • Install ready-made skills and write your own to extend what your agent can do

  • Connect external tools with the Model Context Protocol (MCP) — wire up GitHub and manage issues from Telegram

  • Connect a Telegram bot through the gateway so you can message your agent from your phone, anywhere

  • Monitor your agent in the Hermes web dashboard over a secure SSH tunnel

  • Automate recurring tasks with cron in plain English, and chain jobs into multi-stage pipelines

  • Give your agent a personality and persistent memory that carries across sessions

  • Complete a capstone: a daily briefing your agent generates on its own and texts to you

Requirements: basic command-line comfort, a low-cost VPS (~$6/month), an LLM provider key (OpenAI, Anthropic, or an open source option), and a Telegram account. No prior agent experience needed.

Stop renting your AI. Build one that's always on, answers from your pocket, and lives on a server that's yours.

Who this course is for:

  • Developers and curious hobbyists who've played with ChatGPT, Claude, or Cursor and want to run their own always-on agent on hardware they control
  • Self-hosters and homelab enthusiasts who'd rather deploy and own their AI stack than rent it
  • Indie hackers and solo founders who want a multi-agent "workforce" (coder, tester, deployer) that ships work while they sleep
  • Beginners who want to learn real Linux server administration through a concrete, motivating project — not yet-another generic "intro to Linux"
  • Engineers comfortable on the command line who want a hands-on tour of MCP, agent orchestration, and the operational side of LLMs (systemd, cron pipelines, watchdogs, HTTPS)
  • Anyone who wants a private Telegram or Discord AI assistant they can chat with from anywhere, without paying a SaaS

Report abuse