A global food embedding model for search, dedup, and menu intelligence.
A food-specialized AI API for delivery platforms. One integration handles search, dedup, classification, and menu intelligence across every cuisine.
POST /search
{ "query": "something spicy",
"corpus": ["Chicken 65", "Vanilla Shake", "Peri Peri Fries"] }
// => Chicken 65 (0.91), Peri Peri Fries (0.87)Core Capabilities
Semantic search
"cold coffee" finds Iced Americano, Cold Brew, Frappe. "something spicy" returns Chicken 65 and Peri Peri Fries.
Menu deduplication
Clusters Kadhai Chicken, Karahi Chiken, and Kadai Chkn as one dish. Handles misspellings, transliterations, and promo noise.
Cross-lingual matching
Knows pollo, poulet, Huhn, and chicken are the same protein. Works across 100+ languages and scripts.
Smart cart upsell
Cart has Biryani and Naan? Suggests Raita, Masala Chai, Gulab Jamun. Respects cuisine and dietary preferences.
Cuisine classification
Tags any menu item across 19 cuisines. Pad Thai is Thai. Tacos al Pastor is Mexican. No training required on your end.
Built-in preprocessing
Strips prices, promo text, sizes. "**NEW** 50% OFF Chiken Biryani (Serves 2)" becomes "Chicken Biryani".
Dietary conflict detection
Prevents merging Chicken Burger with Veg Burger. Detects veg/non-veg and specific protein conflicts automatically.
Menu health reports
Quality score 0-100 for any menu. Flags duplicates, noisy text, non-food items like napkins and lighters.
Benchmark Results
20 food-domain benchmarks at 384 dimensions
Retrieval & Matching
Whether the model understands food semantics across cuisines and languages.
Menu Item Deduplication
dish-embed vs the best open-source alternative on food deduplication benchmarks.
Search Ranking
dish-embed vs general-purpose embedding models on food search relevance.
Classification & Robustness
Cuisine classification accuracy and food-domain robustness signals.
Have a question?
Drop a quick note.