Years of user context, in minutes

Turn the camera roll into on-device, queryable context, such as places, experiences, and preferences. No cloud processing required.

Reference device: iPhone 15 Pro (A17 Pro, 16-core Neural Engine, 8 GB RAM), iOS 17+.

Loading the demo…

On-device performance

Reference device
iPhone 15 ProA17 Pro, 16-core Neural Engine, 8 GB RAM.
Target runtime
iOS 18+Apple Vision and Core ML.
Initial processing target
10–20 min / 10,000 photos
Peak memory
400–800 MiB
Battery use target
1–2% / 10,000 photos

One context layer, many apps.

Personal AI

Personal context to recall, organize, and build on.

Events

Recommend events around the experiences they already enjoy.

See it in action

Video editors

The right clips, already connected to their moments.

Travel

Recommendations grounded in places they’ve explored.

Context, ready to use.

Explore the developer docs
Swift
// Existing internal search service
let result = try await searchService
    .searchWithMoments(
        "coffee and pastries",
        limit: 60
    )

// result.hits
// result.momentCandidates

SDK access is available by request.

Your users’ camera rolls stay theirs.

Camera roll
On-device intelligenceVision, text, place & time
Your app

Indexing and semantic search run locally. Optional place and event enrichment use external services; cloud reasoning requires separate consent.

Build with context.

Get API access