FoodHalo

You scan a barcode and see a visual verdict personalised to your goals, allergies, diets, dislikes and values. Each verdict shows an evidence line with how many signals the score rests on.

If a product conflicts with an allergy you declared, a full-screen alert appears. When allergen data is missing, FoodHalo flags it instead of passing the product through. A coach answers questions about what you scan, using product data it looks up. FoodHalo uses wellness wording only and makes no medical claims.

• Scan. Barcodes, QR codes and GS1 Digital Link, plus meals, menus and recipe links.
• Preference Halo. Your goals, allergies, diets, dislikes and values shape every verdict.
• Allergen rule. Declared allergens are never silently passed, and incomplete data is flagged.
• Track. Hydration, caffeine, fasting and meal history.
• Ask the coach. Questions are answered from product data the coach looks up.

A coach that has to look things up

The coach runs through an Anthropic-compatible gateway and must use tools against the product database. The model is not allowed to invent nutrition figures: it writes the sentence, and the data supplies the facts.

One scoring package, two places

The fit score is a pure TypeScript package that runs on the server and on the phone, so verdicts work offline from a local product cache. The design target is a verdict in under two seconds from scan.

Open data, imported on a schedule

Product data comes from an Open Food Facts mirror and USDA FoodData Central. Scheduled imports bring in recall notices and Swiss and European food-composition tables. Open Food Facts data is licensed under the ODbL, which requires attribution wherever its data is shown.

FoodHalo is for anyone who shops with a goal, an allergy or a set of values in mind: people managing allergies, eating for performance or health, or avoiding what they dislike. Scan a product in the aisle, get a verdict that fits you, and ask the coach what you need to know.