Keepers scores, blink-checks, groups bursts and shortlists an entire wedding or portrait session — running on the iPhone itself, straight off your USB-C card.
That session weighs 60–150 GB. Uploading it from a phone is not slow — it's impossible in any reasonable time, on any reasonable plan. So Keepers doesn't try. It culls in place off your card, using the embedded JPEG previews and Apple's Vision framework on the Neural Engine.
Wedding and portrait agreements routinely ban storing images on third-party servers. On-device culling keeps you compliant by default — there's no server to leak.
Competitors pay for cloud compute and pass it on. Keepers runs entirely on the silicon you already carry, so there's no metered server cost baked into the price.
Every RAW file ships a full-res JPEG preview. Keepers scores those on the Neural Engine — fast enough to scan thousands of frames without decoding every RAW.
Five jobs, all on the iPhone: score every frame, kill the duplicates, surface the keepers, learn your taste, and hand stars straight back to your editor.
Each shot gets a composite score from real signals — focus sharpness, eye-open / blink detection, face capture quality, and an aesthetic model. Thermal-aware batching scans thousands of RAW frames in minutes without cooking your phone.
Near-duplicate grouping collapses a 12-frame burst to the single sharpest, best-expression keeper.
Every accept and reject trains a personalization model on-device. Over a few sessions, the shortlist starts to look like the selects you'd have made by hand — your eye, scaled.
Stars, picks and color labels export as sidecar XMP straight back into Lightroom Classic or Capture One. Nothing is baked, nothing is destroyed — you finish exactly where you already edit.
Connect your USB-C card reader or SSD straight to the iPhone. Keepers reads RAW frames in place — no import, no copy.
Thermal-aware batch scoring runs thousands of frames on the Neural Engine — sharpness, blinks, faces and aesthetics, fast.
Bursts collapse, duplicates merge, keepers rise to the top. Review the shortlist and adjust — it learns from every call.
Stars and labels write out as lossless XMP. Open Lightroom Classic or Capture One and your selects are already there.
Photographers already pay $120–540/yr for desktop AI culling. Keepers is mobile-native and on-device — so it can be priced like a tool you own, not a server you rent.
Try the engine on a real session before you commit.
No card required
The full culling engine for working photographers.
or $7.99/mo · billed via App Store
One payment. Yours for good, no renewals.
Best value over two years
Never. Keepers runs entirely on the iPhone using Apple's Vision framework on the Neural Engine. Your photo bytes stay on your card and your device — 0 bytes are uploaded. There is no Keepers server holding your images, which is exactly why it works within client contracts that forbid uploading wedding or portrait files.
RAW from the major camera makers (Canon, Sony, Nikon, Fujifilm and more) plus JPEG. Keepers reads the full-res JPEG preview embedded in each RAW for fast scoring, so it scans thousands of frames without decoding every file.
It doesn't move the session — it reads it in place off your USB-C card or SSD. Scoring runs on the embedded previews via the Neural Engine, with thermal-aware batching so a 3,000-frame session finishes in minutes without overheating the phone.
Yes. Stars, picks and color labels export as standard sidecar XMP, lossless and non-destructive. Import the folder into Lightroom Classic or Capture One and your selects and ratings are already applied.
Those tools proved photographers will pay for AI culling — but they're desktop-only and lean on the cloud. Keepers is mobile-native and fully on-device: cull straight from the card on your phone, between the ceremony and the reception, with nothing uploaded and no server subscription.
On-device AI culling for working photographers. No desktop, no upload, no server bill — just your iPhone, your card, and a finished shortlist.