Your media server now orders its own refills

What it does

SuggestArr sits between your media server and your request manager. On a cron schedule you control from its web UI, it pulls recently watched content from Jellyfin, Plex, or Emby, finds similar titles via the TMDb API, and submits download requests to Seer automatically. The result: your library keeps restocking itself based on actual viewing habits, with no one clicking “request” ever again.

The interesting bit

The optional (beta) AI layer works with any OpenAI-compatible endpoint — including a local Ollama model — so recommendations can run entirely on your own hardware. If the LLM is down or returns nothing, it falls back to the plain TMDb algorithm without drama. There’s also an AI Search tab that turns queries like “80s sci-fi with practical effects” into structured TMDb filters, ranked with a rationale per pick.

Key highlights

• Supports Jellyfin, Plex, and Emby as watch-history sources; requests go to Seer
• Any OpenAI-compatible LLM works: OpenAI, Ollama, OpenRouter, LiteLLM proxy
• Web UI for config, cron scheduling, user selection, and real-time log filtering
• Can skip content already streaming in your country, and exclude titles you’ve already watched
• SQLite by default, with optional PostgreSQL/MySQL backends

Caveats

• The AI recommendations and AI Search are explicitly beta features, gated behind an “enable beta features” checkbox
• Only local Seer users are supported for per-user requests — no SSO/remote accounts yet
• Requires a TMDb API key and a configured Seer instance; it’s glue between services, not a standalone tool

Verdict

Worth a look if you run Jellyfin/Plex/Emby plus Seer and want a self-feeding library. If you don’t already have that stack, there’s nothing here for you.