Semantic search
Find anything in plain language — "sunset on the coast," "red dress, indoors." No tags required; the model understands the pixels.
Private, local AI media library
Private, local AI for your whole library: semantic search, faces, places, video scenes, captions. No cloud, ever.
Free & open source · macOS · Windows · Linux
Everything, understood
Find anything in plain language — "sunset on the coast," "red dress, indoors." No tags required; the model understands the pixels.
Group every shot of a person locally. Face vectors never leave your machine — no cloud face database, ever.
Cluster by location and time into trips and moments, from EXIF you already have — no account, no map provider phoning home.
Split clips into scenes, pull keyframes, and transcribe speech so video is as searchable as your stills.
Every asset gets structured tags and a written caption from local vision models — browsable, filterable, yours.
Learns your keep / reject taste over time and surfaces the frames you'd actually pick. It adapts to you, not the other way around.
How it works
Pick folders of photos and video. Nothing is uploaded — indexing runs on your machine.
Local AI tags, captions, finds faces and places, and splits video into scenes — all offline.
Ask in plain language, filter by person or place, and let it learn what you keep.
Your media stays on disk. There's no server to trust, no account to delete, no breach to fear — because we never receive your library in the first place.
Pricing
Download
First run downloads ~3.5 GB of models, then works fully offline.
FAQ