Sundrop Turns Years of Baby Photos Into Weekly Recaps, Processed on the iPhone
Sundrop, a new iPhone app from the team behind the read-it-later app Matter, turns a parent's accumulated baby photos into weekly recaps, highlights and time-lapses, and it does that processing on the phone. TechCrunch reported the app on November 10, 2026. Ben Springwater, Sky Lan and Rob Mackenzie developed it.
The app scans the pictures in a parent's iPhone Camera Roll, skips screenshots and duplicates, and builds its output from what remains. Image analysis runs through Apple's Vision framework, which detects faces and scenes, and through on-device Core ML models, which are machine learning models that execute on the phone's own chip. Photos are not uploaded to a cloud server or to a hosted AI service operated by Sundrop.
Once a week, Sundrop delivers a recap as a Lock Screen notification, arriving every Sunday. The app also generates scenes, such as placing a child's photo on a magazine cover or a rookie card. It can place two siblings side by side at the same age, and it can assemble a time-lapse series or video of a child growing up.
Two parents can link their accounts and sync shared photos through Apple's iCloud service. According to the report, pictures move between the parents' devices through iCloud without passing through servers run by Sundrop. The report also states that Sundrop does not use generative AI, meaning models that synthesize new images or text, to create its scenes or highlights.
The app costs $39.99 per year and includes a week-long free trial. The annual plan supports Family Sharing, so only one parent needs to buy it for the household to use it.
The broader context here is how the work is divided. Sundrop's vision and classification run locally on Apple's frameworks, and cross-device sync rides on iCloud. The privacy picture therefore rests on the on-device processing Sundrop describes and on how iCloud handles the synced library. Neither depends on a Sundrop-hosted pipeline, which is a different trust model from many consumer photo products. The headline's phrase "without AI" should be read narrowly. Scene and highlight selection rely on Core ML and Vision, which are machine learning systems. What is absent is generative AI.
For developers, the more instructive detail may be the product scope. A baby-photo archive is a bounded, personal dataset with a clear end state, which makes it a useful test of how much a phone can do with its own silicon. Running duplicate filtering, face and scene analysis, and rendering locally avoids a server bill for every family's library.
In my view, the most useful thing about the app is the boundary it draws. A family that has photographed a child almost daily for years holds a large and sensitive collection, and the product's answer to where that collection lives is the phone and the parents' own iCloud accounts. Whether that answer satisfies a given parent will depend on how much they trust iCloud and on whether they want the retrospective at all. For those who do, the result is a summary that arrives as an ordinary Sunday notification, plus a time-lapse of a child growing up, built without the photos leaving the household's devices, for $39.99 per year.
