Runs where you say it runs
A mini PC, an old desktop, a VM, a NAS — if it runs Docker, it can run Josi. Your conversations and memory live on your disk, not on a stranger’s cloud.
Personal AI assistant · Self-hosted & free
Josi is a personal AI assistant you run yourself. Install it on your own hardware — a home server, a mini PC, a box in the closet — and talk to it from every device you own. Your assistant, your machine, your rules.
No ads. No sale of your data. Revocable per-device access
Josi CE · Community Edition
Josi CE is the self-hosted heart of Josi — free to run, built for people who like owning their own stack. If you have ever kept a homelab alive, you already know how to run it.
A mini PC, an old desktop, a VM, a NAS — if it runs Docker, it can run Josi. Your conversations and memory live on your disk, not on a stranger’s cloud.
Install is a docker compose up, not a weekend project. The install guide walks through it step by step.
Point it at the model provider you trust — or a local one. Josi is the assistant around the model, not a lock-in to somebody’s API.
The CE source repository is private while we finish the open beta. No fake “star us on GitHub” buttons in the meantime — when it opens, the link will be right here.
Josi apps · Coming soon
The Josi apps connect to your own server (or a hosted one) and keep one continuous conversation across all of them. Here is exactly where each platform stands — no fake download buttons.
Submitted to Google Play and currently in review.
In beta on TestFlight with early testers.
Desktop apps are in active development.
Want in before the stores? Request early access and tell us what you run.
What Josi does
Josi keeps your conversations, context, and connected tools available when you change devices — without asking you to start over or hand over more access than you meant to give.
Your history follows you across supported devices, so changing screens does not mean starting over. Pick up the same thread on a phone at the door and a laptop at the desk.
Start a voice conversation and keep going. Speech recognition runs only while you are in a voice session, and what you say becomes part of the same thread.
Connect Josi to supported personal-agent services — self-hosted or third-party — while the interface stays calm, familiar, and unmistakably yours.
Each device can receive its own revocable credential, held in operating-system secure storage. One lost laptop does not have to compromise every session.
Send an image, a short video, or a document straight into the conversation when a screenshot explains it faster than a paragraph.
Ask Josi to hold on to a commitment and it comes back at the right moment, with time-sensitive notifications on mobile.
How it works
Run Josi CE on your own hardware with Docker Compose — or wait for the hosted option if self-hosting isn’t your thing.
A one-time setup code creates a signed identity for each device. The credential stays in the device’s secure storage, and you can revoke it on its own.
Type or speak. Your assistant answers with the context it already has, and the thread stays in sync on the rest of your devices.
Trust and privacy
Private beta. Josi is under active development. Security before growth theater: reliability and honest behavior come before sign-up numbers.
Conversation content and credentials are not sold. Content you send is used to deliver the service you asked for.
Josi can connect to self-hosted or third-party agent services you select. Content sent to those services is governed by their terms and privacy practices.
Request deletion of your account and associated data at any time. Read the deletion instructions.
Full detail lives in the Privacy Policy, Terms of Service, and the Google data disclosure for optional Gmail and Calendar connections. Note that AI output may be inaccurate or incomplete — verify anything consequential.
Early access
Josi is in private beta. Leave your name and email — and if you feel like it, tell us what you’d run it on. A person reads these, not a funnel.