FAQ

No. PawCheck gives general product-safety guidance based on label contents, published ingredient knowledge and recall data. It's a smart second opinion in the store aisle — not a diagnosis, prescription or treatment plan. For anything medical, your veterinarian is the only right answer, and every screen in the app says so.

Yes — that's a core design guarantee. Scores are computed by a fixed, versioned rules engine, not generated by AI. The same product scanned for the same pet produces the identical number every time. Scores only change if your pet's profile changes or we publish an updated rules version — and the app tells you when that happens.

Every scan is checked for recalls in two ways: against a curated list of FDA animal & veterinary recalls, and with a live check that uses Google's Gemini AI with Google Search to look for recall notices about that exact product — FDA announcements, manufacturer statements and reputable news. A recall only counts if it names that brand and product line, and anything found is shown with its source. No check is perfect, so if you're worried about a product, the FDA's recall list is the authority.

No — PawCheck reads the label itself with vision AI, so it works on products with worn, missing or regional barcodes, and on things barcode databases don't cover well (like toys and chews). Just get the front label in frame.

Label photos are sent securely to our scanning service, which uses Google's Gemini AI to read the label. They're used only to identify the product and aren't stored on our servers — a small thumbnail stays on your phone in your scan history. We don't sell your data. See the Privacy Policy for the full picture.

PawCheck is a subscription with a free trial — start the trial, scan as much as you like, and cancel in iOS Settings at least 24 hours before the trial ends if it's not for you. Exact pricing is shown in the app, in your local currency, before any charge.

Not yet — PawCheck is iPhone only for now. Android is on the roadmap.

Tap "Not right?" on any result. Low-confidence reads already ask you to retake the photo rather than guessing, and your reports feed a review queue that keeps identification improving.