Building the Dish Intelligence Infrastructure for Dining
Remembite is a dish-level intelligence platform that captures structured behavioral data from restaurant visits — dish reactions, flavor preferences, and repeat behavior — to help users make confident dining decisions in under 15 seconds.
- Remember what you ordered and loved (or didn't) at every restaurant
- Discover community-favorite dishes backed by aggregated reactions
- Predict dish compatibility using AI-powered personal taste modeling
A three-layer intelligence system:
| Layer | Purpose | Features |
|---|---|---|
| Utility | Core tracking | Menu OCR, one-tap reactions, ratings, notes & photos |
| Community Data | Collective intelligence | Aggregated dish reactions, edit governance, structured dish database |
| AI Intelligence | Personalization | LLM classification, Bayesian flavor modeling, taste vectors, compatibility predictions |
- Frontend: Flutter (offline-first)
- Backend: Rust (Actix/Axum)
- Database: PostgreSQL
- AI: LLM-based structured classification with Bayesian community correction
- Phase 1 — Utility core (OCR, reactions, ratings)
- Phase 2 — Governance & data integrity (edit suggestions, moderation, access control)
- Phase 3 — AI layer (taste vectors, compatibility predictions, confidence-gated predictions)
All rights reserved.