FoodyLLM — AI Copilot for Food Nutrition Intelligence
Overview
FoodyLLM is an AI tool designed to analyze and understand food information from text such as recipes, product descriptions, or ingredient lists. It can estimate the nutritional content of recipes (for example fat, carbohydrates, protein, sugar, and salt), predict nutrition labels like the UK FSA traffic-light system, and convert food descriptions into standardized, structured data. This helps companies work with food data more easily, support nutrition analysis, and enable better digital food and health applications. FoodyLLM is built on a transformer-based LLM architecture fine-tuned in a multi-task learning setup to jointly handle macronutrient estimation, FSA traffic-light profiling, and semantic food data mapping. The model leverages food ontologies and nutritional databases during training to ground its predictions.
Outcome
Given a recipe's ingredients and quantities (and optionally its title), the model outputs estimated macronutrient values (fat, carbohydrates, protein, salt, sugar), FSA traffic-light nutritional labels (low/medium/high) per nutrient, and semantically annotated food data aligned with established food ontologies for interoperability with FAIR data repositories and health information systems.
Service Card
| Category | AI products |
| Domain | Personalized Medicine, Nutritional Informatics, Food Science |
| Vertical | Health and Biotechnology |
| Target users | Nutritionists, Food Scientists, Health Data Professionals, App Developers |
| Service type | Model |
| Outputs | Macronutrient estimates (fat, carbohydrates, protein, salt, sugar); FSA traffic-light labels (low/medium/high); semantically annotated food data mapped to food ontologies. |
| Requirements | GPU recommended for inference at scale; recipe input as ingredients with quantities (recipe title optional). |
| Limitations | Model performance is optimised for food and recipe domain inputs; accuracy of macronutrient estimation depends on the completeness and specificity of provided ingredient quantities; FSA profiling is based on UK FSA guidelines and may not align with other national nutritional standards. |
| Availability | Q4/2026 |
| Licensing | MIT and CC BY-NC-SA 4.0 |
| Visual content | assets/service-visuals/foodyllm-ai-copilot-for-food-nutrition-intelligence.jpg |
| Visual caption |