Transparency, before the number

How the score works.

A guide to what the app assesses, what the numbers mean, and where they fall short.

Where the information comes from

The app uses photos of product labels or product information associated with a barcode. Barcode product data comes from Open Food Facts (ODbL). Coverage and freshness vary. AI reads the available ingredients and nutrition information to generate an analysis. Previously generated product analyses may be cached and reused.

What the AI is asked to assess

The analysis includes an overall 0–100 score, individual ingredient explanations, and sub-scores called “Additive Safety”, “Nutritional Value”, and “Naturalness”. These are AI judgments, not independent laboratory measurements. Category guidance differs for food, cosmetics, cleaning products, vitamins, and supplements. Athlete mode adds sports-nutrition context and can change the assessment.

How to read the scale

The app’s scoring instructions broadly associate higher ranges with fewer concerns in the available ingredient and nutrition information, and lower ranges with more concerns. These are model instructions, not validated health thresholds. A one-point difference should not be treated as a meaningful scientific distinction.

There is no published fixed formula that combines the sub-scores into the overall number. The AI produces the overall score using the label information and scoring guidance. The number is neither a percentage of healthiness nor the probability that a product is safe.

Nutrition and ingredient limitations

The model is instructed to report nutrition values from a visible panel, normalized to 100 g or ml when conversion is possible, and leave unreadable or absent values unknown. It may still make extraction or calculation errors. Ingredient explanations can cover up to 15 prioritized ingredients; they are not guaranteed to reproduce the entire label.

Why results can differ

Incomplete photos, changes in product formulation, database records, AI providers, and model updates can affect results. Cached analyses may reflect an older label. Different product categories and athlete-mode results are not necessarily directly comparable.

What a score cannot tell you

A high score does not establish that a product is suitable for your allergies, medical condition, medication, age, or dietary needs. The app does not test products, diagnose conditions, or replace professional advice. “Naturalness” is a model assessment, not proof of safety.

Found an error?

Use the feedback option in the app, or contact the developer. Reported analyses may be reviewed or regenerated; reports do not make an existing result safe to rely on.

This explanation reflects the app source inspected on 27 October 2026. Features and scoring instructions may change.