Visual Product Search

Visual Product Search finds products by what they look like. Point your iPhone's camera at an item and the app shows the closest match from your catalog, plus the next most likely matches. Everything happens on your phone. An on-device machine learning model turns each photo into an image embedding, and a vector search over a local database finds the products that look most alike. No account, no internet connection, and the app never uploads your photos. KEY FEATURES • Search by photo: take a picture of a product and see the best match, with other likely matches to choose from. • Your own catalog: add products with a name, price, aisle, description and several photos, so each one is recognised from more than one angle. • Sample catalog: add ten sample products with one tap and try a search right away. • Works offline: search keeps working in a stockroom, a warehouse or on a plane. • Private by design: no sign-in, no ads, no analytics, no tracking. Delete a product and its stored photos are deleted with it. HOW IT WORKS Visual Product Search is built by Infinite Red. Photos are embedded with Apple's Vision framework on device, and products are stored and searched with Couchbase Lite's on-device vector search. It shows how on-device AI can make a fast, private, offline product lookup for retail, inventory and field teams. ABOUT INFINITE RED Infinite Red is a team of experienced engineers who learn your business, work as part of your team, and take responsibility for delivering software you can rely on. We help companies decide what to build, take on the hard problems, and put AI to practical use in real projects. We also maintain open-source tools like Ignite and Reactotron and bring developers together at our Chain React conference. Have a project in mind? Email us at hello@infinite.red.