MediaFind transcribes and indexes your media library locally, then lets you find any clip by describing it in plain language — with exact timestamps.
Point MediaFind at a folder. It transcribes, embeds, and indexes everything locally — then opens up every kind of search and recall.
Describe a moment in plain words and get ranked clips with exact timestamps — semantic, not keyword.
Searches the visual content too — CLIP keyframe search plus OCR of on-screen text.
Keyless on-device diarization labels who said what across your library.
Per-file summaries and auto-segmented chapters so long recordings are skimmable.
Ask a question and get a grounded answer that cites the source files and timestamps.
Strictly opt-in, on-device face library — “show me where this person appears.” Never leaves your machine.
Turn any recording into action items, decisions, and key points — with optional local mic capture.
Pull in public media from a URL so it transcribes and becomes searchable like everything else.
Auto categories, your own collections, knowledge map, and highlight reels of how files connect.
Not just a claim — it's verifiable at runtime. The core path uses no account, no API key, and no telemetry.
| What you do | What leaves your machine |
|---|---|
| Core: index, search, ask, export, clip, summaries, meetings, notes & collections | Nothing |
download / capture a public URL you asked for | Only that public URL — keyless, opt-in |
People & Faces (--faces) | Nothing — opt-in, on-device |
| First run of optional ML models | One-time public model download, then fully offline |
No subscription, no account. Buy once, use forever. Sold through Lemon Squeezy.
macOS app, signed & notarized. An iOS companion is on the way.