CarbonTrac

CarbonTrac is an API for food retailers and GLP-1 telehealth providers. We read the loyalty data supermarkets already have, score every product a shopper buys across 42 micronutrients, protein quality, ultra-processing, diversity and glycaemic load alongside carbon footprint, and send personalised swaps back into the retailer’s own app. Nothing to log, scan or download. Diet is the world’s leading risk factor for death and disease, and supermarkets hold the best data on it.

We found that the healthiest basket is also the biggest one. Conventional personalisation sells more of what people already buy and tops out at a 1-2% uplift. We sell people what they’re missing: a deficiency is a category they don’t shop, a pairing is a cross-sell. We recruited 2,819 shoppers to a 12-week beta, nutritional balance improved 19.6%, basket carbon fell 15.5%, basket value grew 6-8%. The same engine is now going into GLP-1 telehealth, where patients eat dramatically less and nobody tracks what they’ve stopped eating.

CarbonTrac is featured in a World Health Organization Regional Office for Europe policy brief on digital interventions for non-communicable disease, now with 27 member state health ministries, and was named a top AI for Good use case of 2026 by the ITU.