Canopy Pantry
A companion app for Canopy, which crowdfunds independent contaminant testing of food. Take a photo of a pantry shelf, let AI vision identify each product, and see it matched against Canopy’s lab results, sorted worst-first, with the tests worth requesting for anything not yet covered.
- Works on
- Native iOS and Android
- Status
- In testing
- Who can open it
- Not in app stores yet
- My role
- Product design
- Made with
- Hova (engineering)
- Built with
- Expo · React Native · Supabase Edge Functions · Claude vision
The screens.
Concept screens with made-up products and people, drawn from the app’s design. Not captures of the shipped app.
The good parts.
One photo, every product
A two-pass vision pipeline finds each product on the shelf, then takes a closer look at each one.
Matched to real lab data
Products are fuzzy-matched to Canopy’s test requests, and severity is computed with the same hazard rules as the website.
Per person, not per pantry
Household members, how much each eats, and projected exposure, with sensitive details asked only when they apply.
Request what’s missing
Each untested product suggests the tests that fit it, with a prefilled request.
Under the hood.
The AI key never ships in the app: photos go to a private bucket and are analysed server-side behind a provider-agnostic interface.