Find a Person · AI Search (beta)
Point FolioSort at a folder, add a few photos of the person, and it finds the rest — the group shots, the ones from across the room, the candids you'd never dig out by hand. It all happens on your computer: no cloud, no account, no face ever leaves your device.
Every shot of your kid, your partner, a close friend is buried across folders, shoots and years — the posed ones and the candids, the close-ups and the group photos where they're half-hidden at the back. Finding them all by hand means scrolling forever and still missing half.
The cloud tools that can find them want your whole library uploaded and your faces scanned on someone else's servers — turning your most personal photos into someone else's data.
Add 1–3 clear photos and Find a Person detects and aligns each face, then compares facial identity — not just "looks similar" — to list every match in the folder, closest first. A match-strength slider tunes how strict it is; you review each one before anything happens.
Then copy the keepers out to a new folder, or move them. On macOS the whole pipeline runs on Apple's on-device Vision plus a Core ML recognition model on the Neural Engine; on Windows it uses local models. Either way, no photo or face is ever uploaded.
How it works
Pick 1–3 clear, front-facing photos of the person.
Select where to search, with or without subfolders.
FolioSort scans on-device and lists matches, closest first, each with a confidence score.
Tune the match strength, review every match, then copy or move the ones you want to a folder.
Why it's private
I had ten years of photos of my daughter scattered across a dozen folders. I gave FolioSort three pictures and it pulled out the ones I'd forgotten — including group shots where she's barely in frame. And it never asked me to upload anything, which is exactly why I'd never used the cloud version.
FolioSort's AI Search runs entirely on your device. Find a Person pulls every photo of someone out of the pile; Find Screenshots sweeps out the ones that aren't really photos at all.
Add a few reference photos and find every shot of that person in a folder — entirely on-device.
This pageSpot the screenshots hiding among your real photos and clear them out in one pass.
Explore Find Screenshots → Also clean upFind exact and visually similar copies — even resized and re-edited ones — and reclaim the space.
Explore Duplicates →FAQ
No. Detection and recognition run entirely on your device — macOS uses Apple's on-device Vision plus a local Core ML model; Windows uses local models. No cloud, no account, and it works offline. No photo or face ever leaves your computer.
Those scan your whole library, often in the cloud. FolioSort is the opposite: you give it a few reference photos and a folder, and it finds that one person locally — nothing is uploaded and you review every match before acting.
It compares facial identity, so it's far better than "looks similar". It's in beta and sorts matches by confidence with a strength slider, and you review each match before copying or moving — so you're always in control. Clear, front-facing reference photos give the best results.
Searching and reviewing are free and show every match. On the free plan you can copy or move 20 photos in total from your matches; Pro removes the limit for a one-time €14.99.
Give FolioSort a few reference photos and a folder, and watch it surface the shots you'd never dig out by hand — with nothing ever leaving your computer.