Eidon AI

Eidon AI is a self-hosted AI assistant for people who want everyday chat and agents that can carry out longer tasks on their own computer or server. A Chief of Staff agent answers directly, delegates work to specialists, or creates an agent for a task. Agents can message each other and keep working on a schedule.

Each agent has its own conversation, memory, files, and browser, plus access to a shared team folder. They can research questions, compare travel options, draft inbox replies, or write and run code in a sandboxed workspace. Research reports include sources. You can watch browser activity live and take over when a login or another step needs your attention.

Permissions let you allow shell commands and MCP tools once, always, or never. Outgoing email and Slack messages are drafted for review before sending. The browser password flow is designed to enter credentials without including them in model prompts or conversation text. You can stop or redirect work, rewind a conversation, or branch from an earlier message.

The everyday chat includes editable memory, folders and search, voice input, file and image attachments, and temporary conversations. Both chat and agents can use web research, image generation, MCP tools, and reusable skills.

Eidon runs in Docker and is free and open source under AGPL-3.0. It supports Ollama and LM Studio for local models, alongside OpenAI, Anthropic, OpenRouter, and compatible APIs. Cloud providers handle model requests when selected; Eidon stores chats on your server and encrypts provider credentials. You can assign a different model to each agent. Multiple users have separate private data, and the web app can be added to a phone's home screen.