One platform that ingests fragmented food data, resolves it into a global knowledge graph, and serves it through developer-first APIs. Built for the AI era — not for the ordering screen.
A continuous pipeline that runs on every menu in the world, 24/7.
Our crawlers and AI agents collect menus from websites, PDFs, images, delivery platforms and POS exports — in every language and format.
Entity resolution maps "Spicy Pork Ramen", "Karai Tonkotsu" and "辣豚骨拉面" to the same dish. Duplicates collapse. Ambiguity disappears.
Every entity is typed, linked and enriched: dishes know their ingredients, cuisines know their dishes, allergens propagate through the graph.
Structured intelligence is exposed through REST APIs, streaming datasets and webhooks — ready for any AI application to consume.
Each layer is independently powerful. Together they are the foundation of food AI.
The extraction brain. Computer vision and NLP agents parse any menu — a blurry photo, a scanned PDF, a 14-language delivery listing — into clean, typed fields: section, item, price, description, modifiers.
See how it worksThe semantic core. A global, continuously evolving graph of restaurants, menus, dishes, ingredients, cuisines, allergens, nutrition and prices — queryable as one connected web of food.
Explore the graphThe developer surface. Predictable, typed REST endpoints for restaurants, dishes, ingredients, nutrition and intelligence — with SDKs, webhooks and a playground.
Read the API docsThe training substrate. Enterprise datasets for fine-tuning food-aware models — clean labels, global coverage, continuous refresh, clear licensing.
Explore datasetsEvery schema, endpoint and dataset is designed for LLM retrieval, embeddings and agents — not retrofitted from a POS screen.
12 languages out of the box, 180+ countries, per-location pricing and cultural cuisine handling baked into the core.
99.99% uptime, sub-50ms p95 reads, streaming sync and SOC 2-aligned security. Built to sit inside your stack, not beside it.
Start with the free tier. Scale to billions of data points. Ask for the enterprise architecture review.
