Carte AI
An iOS menu scanner that translates any menu, explains every dish, and recommends what to order against your own taste profile.
- Swift
- SwiftUI
- iOS 17+
- Supabase
- Claude API
- StoreKit
Project overview
Carte AI is an iOS app that makes ordering off any menu effortless. Photograph a menu and it comes back translated, with plain-language descriptions of every dish and a ranked set of picks chosen against your own taste profile. It is built for travellers eating somewhere unfamiliar, anyone with dietary restrictions, and anyone who has ever stared at a menu with no idea what to order.
It is live on the App Store as a free download in Food & Drink, built and shipped solo end to end: the SwiftUI client, the backend that processes scans, the taste profile model behind the recommendations, and the purchase and credit system that pays for the inference.
Technical features
- Menu scanning pipelineA photographed or uploaded menu is sent to the backend and read by Anthropic's Claude API, which returns a structured list of dishes rather than loose text, so the app can render, rank, and filter them.
- Taste profile matchingAn onboarding quiz captures palate preferences, allergens, and restrictions. Recommendations are scored against that profile and anything that breaks a restriction is kept off the list entirely.
- Translation and explanationMenus in all major languages are translated into the user’s own, and each dish is explained in plain language so an unfamiliar name is not a guess.
- Accounts and scan historySupabase handles authentication, the Postgres database, and image storage, with row-level access control. Scans and their results persist so they can be revisited, and the user can delete any of them.
- Credits and subscriptionsApple processes every transaction and the backend verifies the signed receipt before updating the credit balance and subscription tier. No payment details are ever collected or stored by the app.