Layer 1
Cultural Food Dataset
A curated, continuously expanding database of culturally specific ingredients, dishes and preparation methods, covering West African, South Asian, Caribbean and Middle Eastern cuisines.
How it works
Simple steps. Meaningful change.
Three steps for you. Four intelligent layers underneath.
01
Describe your meal
Type it, photograph it, or pick from a cultural recipe library. No calorie maths.
02
CADDE interprets it
Recognises ingredients, cooking methods and portion norms from your cuisine, not a Western database.
03
Get practical guidance
Suggestions you'll actually use, like swapping white rice for parboiled, or rebalancing your eba portion.
Core innovation
Meet CADDE.
Our Culturally Adaptive Dietary Decision Engine reads cultural meals the way generic nutrition APIs can't: ingredient by ingredient, portion by portion.
- Cultural ingredient mapping
- Personalised substitutions
- Glycaemic impact modelling
- Multi-language input
- Portion adaptation by cuisine
Architecture
Four layers, built to scale with the community using them.
Layer 2
Meal Adaptation Engine
Interprets user-submitted meals and generates evidence-based substitutions and portion adjustments aligned with diabetes dietary guidelines.
Layer 3
Behaviour Learning System
Tracks user choices over time to learn preferences and improve suggestions. A personalisation loop that strengthens with use.
Layer 4
Metabolic Intelligence Layer
Future integration of CGM data and metabolic markers to deliver predictive, real-time food guidance tailored to your glucose response.
Help shape what GlucoSteps recognises first.
Early testers tell us which dishes, swaps and language work in real kitchens. That's the data CADDE learns from.