How to Setup Hermes Agent with OpenRouter (Step-by-Step Guide)

How to Set Up Hermes Agent with Open Router and Telegram

Introduction to Hermes Agent

• The speaker introduces the tutorial on setting up the Hermes agent with Open Router and Telegram for personal AI control.

• Emphasizes the importance of understanding the architecture of Hermes, which is more than just a chatbot; it can evolve into a personal AI assistant that learns and grows over time.

Advantages of Using Hermes

• Highlights that unlike traditional chatbots, Hermes functions as an agent runtime capable of executing tasks rather than merely answering questions.

• Discusses its persistent memory feature, allowing it to remember interactions across sessions and improve its functionality by learning user preferences.

• Notes that Hermes can perform multi-step workflows autonomously, enhancing its utility in various tasks.

System Architecture Overview

• Clarifies that while Hermes is not the AI itself, it operates using an LLM (Large Language Model) from Open Router as its brain.

• Describes how Telegram serves as the interface for user interaction with the agent, while Hermes executes commands based on LLM decisions.

Setting Up Your Environment

• Instructs users to visit the Hermes GitHub page for installation instructions and emphasizes self-hosting options on personal computers or VPS.

• Mentions using Azure VM for demonstration purposes but notes that local installations are also viable.

Installation Process

• Guides through installing the Hermes agent via a one-line installer suitable for Linux/Unix systems; Windows users need WSL (Windows Subsystem for Linux).

• Explains obtaining an API key from Open Router after creating an account, highlighting that usage incurs costs based on requests.

Open Router and API Key Management

Understanding Open Router

• Open Router serves as a centralized platform for selecting various models, functioning like a marketplace for large language models (LLMs). It allows users to manage multiple API keys instead of being limited to one model.

Cost Considerations

• Users should be aware that costs can escalate quickly depending on usage. Light users may spend around $20, while heavy users utilizing expensive models could incur charges of $200 or more.

Model Recommendations

• For cost-effective usage, it is recommended to start with three specific models: Google Gemini 25 Flash, Open AI GPT5 Nano, and Moonshot AI Kimmy K 2.5. These options help avoid rapid depletion of credits.

Configuration Settings

• The configuration process involves setting up the terminal backend (local vs. Docker), max iterations for tasks (initially set at 60 but adjusted to 70 based on advice from ChatGPT), and tool progress mode preferences.

Context Management

• Users can choose how often the context resets—options include daily or never auto-resetting. A persistent memory feature in Hermes allows it to retain information across sessions until it reaches capacity.

Setting Up Telegram Bot Integration

Interaction with Hermes

• The setup includes configuring a Telegram bot token for user interaction with Hermes, which enhances usability compared to terminal-only access.

Creating a Telegram Bot

• To create a bot, users must have a Telegram account linked to their phone number and interact with BotFather on Telegram. This process is straightforward but requires caution regarding privacy.

User ID Configuration

• After creating the bot, it's essential to obtain the user ID using a user info bot so that the newly created bot only receives messages from the user's account.

This structured approach provides clarity on key concepts discussed in the transcript while ensuring easy navigation through timestamps for further exploration of each topic.

Setting Up Hermes and Telegram Integration

Initial Setup Steps

• The speaker emphasizes leaving default settings and proceeding with the setup by clicking enter, indicating that using a local browser is sufficient for this process.

• After initial configurations, the speaker confirms that Telegram and Open Router are set up, along with downloading Hermes. They mention running Hermes in the command line to initiate the terminal user interface (TUI).

Configuring Gateway for Telegram

• The speaker checks the current model configuration of Hermes, confirming it is set to "open router auto," which is essential for proper functionality.

• A critical step involves setting up the gateway; without it, interaction via Telegram won't be possible. The speaker runs Hermes gateway setup and notes their bot shows as not configured due to not creating a new bot yet.

Finalizing Bot Configuration

• The speaker successfully configures their Hermes bot and opts to install the gateway as a systemd service, ensuring it runs in the background on boot. They stress entering your user ID during configuration for security.

• After restarting the gateway service, they test communication with their bot by sending a message and receiving an automated response.

Setting Up Important Files

• The next steps involve navigating into the Hermes directory where important files like API keys (av file) and soul.md are located.

• The soul.md file defines Hermes' persona—its personality and tone—which can significantly enhance its usefulness based on how it's described.

Enhancing Interaction with Memory Files

• The speaker mentions customizing Hermes' persona using insights from ChatGPT instead of relying on built-in options, highlighting how personalization improves interaction quality.

• They explain that interacting with Hermes will generate a memory.md file containing important information it learns about you over time.

Utilizing Skills Directory

• Users can create or modify a user.md file to provide context about themselves to improve interactions further. This allows Hermes to understand user preferences better.

• The skills directory contains various capabilities of Hermes organized by categories such as productivity tools and web scraping. It can also develop new skills based on user interactions.

By following these structured steps, users can effectively set up their personal assistant using Hermes while integrating it seamlessly with Telegram for enhanced functionality.

Understanding the Hermes Agent

Overview of the Hermes Agent's Capabilities

• The Hermes agent is described as an autonomous system, not just a chatbot, capable of proactive tasks and learning about the user over time.

• Users can connect various tools such as calendars and APIs to enhance functionality, allowing the agent to manage tasks automatically rather than requiring manual input.

• The system can develop reusable skills through repeated actions, demonstrating a form of self-awareness in creating its own behaviors based on user interactions.

• Caution is advised regarding security; users should be mindful of what access they grant to the agent, especially concerning sensitive information like production emails or API keys.

• The video concludes with encouragement for users to experiment with their Hermes agents and share experiences, hinting at potential future content focused on tool integration and automation.