Mervin Praison

Hermes Agent Locally: No Cloud, No Limits

Run Hermes Agent 100% locally on your own computer with Ollama — free, private, and connected to Telegram so you can hit it from anywhere. Self improving agent. In this walkthrough I install one of the most popular open-source agent platforms (100k+ GitHub stars), wire it up to a local Gemma model, and chat with it from my phone in under 6 minutes. Links: Ollama: https://ollama.com https://docs.ollama.com/integrations/hermes https://hermes-agent.nousresearch.com/ What's inside: Install Ollama with a one-liner Pull Gemma 4 E2B (fast enough for a 34GB Mac) Install Hermes Agent with ollama launch hermes Create a Telegram bot via @BotFather Lock the bot to just you using @userinfobot Install the gateway so the agent stays online Send your first message from Telegram and watch it respond Why local: Your data never leaves the machine Zero API cost, no rate limits Works offline Full control over model and config Trade-offs to know: Local models won't match frontier cloud models on the hardest tasks You'll want enough RAM to run the model comfortably (I'm on 34GB) The gateway needs to stay running for Telegram to keep responding Chapters: 00:00 What we're building 00:45 Install Ollama 01:18 Download Gemma 4 E2B 02:04 Install Hermes Agent 02:43 Create a Telegram bot with @BotFather 03:50 Restrict the bot with @userinfobot 04:21 Install the gateway 04:26 First test in the chat interface 04:38 Chatting from Telegram 05:12 Wrap up My channel: https://youtube.com/@MervinPraison If this helped, hit subscribe and drop a comment with the first task you're going to give your local agent. #Ollama #LocalLLM #AIAgents #Telegram #OpenSource #SelfHosted This video shows how to set up the Hermes Agent, an open-source "ai agent" platform, completely "free ai" and locally for private data management. The agent is integrated with Telegram, allowing you to chat with your "telegram bot" and assign tasks. We'll walk through the "setup" process, including configuring the gateway and messaging platforms, and demonstrate the agent's available tools and skills for a truly "local ai" experience.