Ask, don't search
"Is the new season of that show here yet?" Gary checks your library, your requests and what is still downloading, and answers in one reply.
Open source, self-hosted, opinionated about movies
Meet Gary,
your media clerk.
Gary is an AI media concierge that sits on top of the stack you already run. Ask him what is worth watching, request it in plain language, and he will tell you when it is ready. He works with Plex, Seerr, Sonarr, Radarr and Tautulli, and he never changes anything without showing you first.
Confirm, then done
Requests go through Seerr with an exact preview you approve. Per-person quotas and separate histories keep a shared household tidy.
He tells you
Gary messages you when a request is ready to watch, and suggests what to watch next from your own viewing history.
Talk to him where you already are
Gary's chat channels are adapters over one shared chat library, so a new channel is a small, self-contained addition rather than a rewrite.
- Web chat built in, with nothing extra to set up.
- Telegram with a single bot token.
- WhatsApp for the patient, through the official Cloud API.
Bring your own brain
Point Gary at an OpenAI key, or at any OpenAI-compatible endpoint you run yourself, such as Ollama, LM Studio, vLLM or OpenRouter.
- Your stack stays yours. Gary talks to your services over your own network, and your API keys never leave it.
- No stack yet? An opinionated Baymax media stack is planned as an optional companion, not a requirement.
Try it
Gary has been the concierge for a small private group for months. The first release is out for anyone who wants to run him for their own household. It needs Docker, Seerr and an OpenAI key or any OpenAI-compatible model. Follow the quickstart in the README. Expect rough edges, and file an issue when you hit one.
What comes later
- A guided first-run setup in the web UI.
- More chat adapters, starting with Discord.
- A hosted, one-click option for people who would rather not run another container.