May 23, 2026 · Pebble Studio
Why Pebble runs entirely on your device
A cleaner app needs access to your photos. That is unavoidable.
What it does with that access is the whole story. Most cleaners upload your photos to a server, run a model there, send back the result, and “forget” to mention this in the privacy label. Some apps process on-device but still send hashes, filenames, or thumbnails to analytics. Both are common. Both are wrong.
Pebble takes a stricter line: nothing about your photos leaves your phone. No upload. No hash. No filename. No thumbnail. No “anonymous” identifier that could be joined later. The duplicate detection, the blur classifier, the burst grouping — all of it runs in the same process as the app, on the same device, against your local photo library.
We pay a real cost for this. Server-side ML models are bigger and faster. Server-side analytics make us smarter about what to improve. We are deliberately choosing the slower, harder, less data-rich path because the trust we want from you is worth more than the convenience we lose.
If you ever want to verify this, install the app and put your phone in airplane mode. The cleanup still works. That is the test.
— Pebble Studio