Odysseus is a self-hosted AI workspace that brings chat, agents, research, documents, email, notes, tasks, calendar, and local model workflows into one application. It is deployed with Docker Compose and served on a local port, and each piece runs against whatever model endpoints you point it at, with no telemetry and optional external integrations you turn on yourself.
Chat sessions can be given tools — bash, files, web, and memory — plus any MCP (Model Context Protocol) server you connect, toggled per tool, so an agent can plan, call tools, and work through a task rather than only answer. A Cookbook surface makes hardware-aware model recommendations and handles downloading and serving models across your machines, and the assistant writes and reuses its own skills.
Features
- Chat and agents: local or API models with tools, MCP servers, files, shell access, skills, and memory
- Cookbook: hardware-aware model recommendations, downloads, and one-click serving
- Deep Research: multi-step web research that gathers and reads sources, then writes a cited report
- Compare: one prompt sent to several models at once, answered side by side and synthesised
- Documents: a writing-first editor with AI edits and suggestions, Markdown, HTML, CSV, and syntax highlighting
- Email: an IMAP/SMTP inbox with triage, tags, summaries, reminders, and style-matched reply drafts
- Notes, tasks, and calendar: reminders, todos, scheduled agent tasks, and CalDAV sync
- Memory: persistent memory the assistant builds up and recalls across conversations
- Extras: image gallery and editor, themes, uploads, web search, presets, sessions, and two-factor authentication
