Odysseus (AI)
Canonical version: Odysseus (AI).
Odysseus is a self-hosted AI workspace that puts chat, autonomous agents, research, documents, email, notes, calendar, and local model serving behind one local-first interface, with no telemetry. Python + JavaScript, AGPL-3.0-or-later.
The pitch: instead of stitching together a chat app, a research tool, an email client, and a model server, you run one workspace on your own hardware and point it at whatever model endpoints you choose. It sits in the same self-hosted, local-first AI-assistant space as OpenClaw and Hermes, but its shape is a workspace (productivity surfaces around the agent) rather than a messaging gateway.
Subsystems
For atomic deep-dives:
- Odysseus Cookbook — hardware-aware model recommendations, downloads, and one-click serving across 270+ models.
- Odysseus Deep Research — multi-step web research with source reading and cited reports.
- Odysseus Email Assistant — IMAP/SMTP triage, summaries, style-matched reply drafting.
- Odysseus Compare — blind side-by-side model testing and synthesis.
Plus, in the core workspace:
- Chat & Agents. Multi-turn chat and autonomous agents that plan, invoke tools, and work through tasks. Runs local or API models.
- Tools & Model Context Protocol (MCP). Built-in bash, files, web, and memory tools, plus any MCP server, each toggleable per tool.
- Memory & Skills. Persistent context across conversations; the assistant autonomously writes and refines reusable skills over time.
- Documents. Writing-first editor with AI edits and suggestions; Markdown, HTML, CSV, syntax highlighting.
- Notes, Tasks & Calendar. Reminders, todos, scheduled agent tasks, CalDAV sync.
- Image gallery & editor, themes.
Deployment
Quick start is Docker Compose; the UI lands on localhost:7000. GPU support ships in NVIDIA and AMD variants; native install is also supported. Session and preset management plus 2FA are built in.
Branch model: dev is the default branch and gets the newest changes; main is the more curated release.
Security posture
The project's own guidance: keep auth enabled, keep private data out of Git, and never expose raw model or service ports publicly.
References
Related
- OpenClaw
- Hermes
- AI Agents
- Model Context Protocol (MCP)
- Odysseus Cookbook
- Odysseus Deep Research
- Odysseus Email Assistant
- Odysseus Compare
About Sébastien
Ready to get to the next level?
Found this valuable? Share it with someone who needs it.