PicGent uses semantic search and OCR (text recognition) to help you locate pictures by what they contain. Just describe what you remember or search for words that appear in a photo, screenshot, or document — and get results in seconds.
Currently available on iOS — Android version coming soon.

Search by what you remember, or by the text inside the photo.
Natural-language search across scenes, objects, and context. Try: “a latte with leaf-shaped foam art on a wooden cafe table,” “a crowded night market with colourful lanterns and food stalls.”
OCR finds words inside your photos — screenshots, signs, menus, documents, notes, and receipts. Try: “ppt of 50 AI companies of 2025,” “receipt from Starbucks.”
Three steps. You control which photos the app sees.
Pick Limited Photos and choose exactly which images the app can use, or grant full-library access — your call.
The app securely sends your selected photos to our servers to compute embeddings, then discards the photos on the servers. Embeddings come back to your device.
Describe what you want → search runs on your device → pick the results → save to a new album or share.
We process only what’s needed for search, and nothing more.
Grant Limited Photos access and pick exactly which images the app can use — at any time.
Selected photos are sent over TLS to compute embeddings, then permanently deleted from our servers. No photo or image data is retained.
Your photos are not used to identify you, profile you, train models, or sell to third parties.
Quick answers before you download.
No. Selected photos are sent over TLS to compute embeddings, then permanently deleted from our servers. No photo or image data is retained.
Yes. Grant Limited Photos access in iOS and pick exactly which images PicGent can use — or grant full-library access if you prefer.
Natural-language descriptions of scenes, objects, or context (e.g. “A starfish on the beach”), and text inside photos via OCR (e.g. “ppt of top 50 AI companies of 2025”). Vague inputs still work.
English and Chinese.