Your Own AI Workspace, On Your Own Machine
Odysseus is a free, open-source workspace that runs AI assistants, email triage, and research agents entirely on your own computer — no subscriptions, no data sharing.
What this is
Imagine if ChatGPT lived entirely on your computer. No subscription. No company logging your questions. No data leaving your office. That's the idea behind Odysseus — a free, open-source project that just launched this week and is already one of the most-watched new tools on GitHub.
You set it up once, and from then on you have a full AI workspace: a chat interface, an assistant that can search the web, read and write files, and even sort through your inbox automatically. It comes with over 270 AI models you can download in one click — the software looks at your hardware and recommends what will actually run well on your machine.
Why it's interesting
Most AI tools today are services — you're renting access to someone else's computer, and your conversations go through their servers. That's fine for many things. But if you run a law firm, a clinic, a design studio handling client briefs, or frankly any business where confidentiality matters, that arrangement deserves a second thought.
Odysseus flips it around. Everything stays local. The email triage feature alone is worth watching — it connects to your existing inbox and uses AI to sort, flag, and summarise. No third party touches your messages.
It's very new and still a bit rough around the edges. But the timing is good to pay attention.
Worth thinking about
What conversations or documents in your business would you rather keep completely off external servers?
Quick glossary
Self-hosted — software you run on your own computer or server instead of accessing through a website someone else controls.
Open-source — the code is public and free. Anyone can inspect it, use it, or improve it. The opposite of a black box.
AI agent — an AI that doesn't just answer questions but can take actions: browsing the web, reading files, sending emails on your behalf.
MCP (Model Context Protocol) — a standard that lets different AI tools talk to each other, like a shared language between apps.