Two people, one scan, two different answers

Scan the same granola bar as a friend without any profile active and you'll both see the same base score. Turn on a Celiac profile and scan it again, and a gluten-containing ingredient can flip your personal verdict to avoid — even though the generic score didn't move.

That split is deliberate. A single number can't tell you whether a product is right for your body; it can only tell you how the product looks in general. Ingredo.app shows both numbers, side by side, with the reasoning behind each one visible.

Stage one: the base score

For food, the base score reuses Nutri-Score (nutrition), NOVA (processing level), Eco-Score (environmental impact) and additive/allergen analysis already computed by Open Food Facts — we don't re-derive these from scratch, we cite them.

For cosmetics, the base score starts from Open Beauty Facts ingredient data and adds our own cited tags for what that source doesn't cover — comedogenic ratings and the EU Annex III fragrance allergen list, for example.

Unknown ingredients are never guessed at. They're marked "insufficient data," and if an offline AI enrichment step proposed a tag rather than a verified source confirming it, the score is marked AI-influenced with the contributing estimates listed for you to inspect.

Stage two: your personal verdict

Your profile is the set of condition and diet modules you've turned on — Celiac, Low-FODMAP for IBS, lactose intolerance, vegan, nut allergy, sensitive skin, fragrance allergy, acne-prone. Each active module reacts to specific ingredient tags.

Every tag maps to one of three flags: avoid (a hard block — gluten for a Celiac profile), caution (a soft penalty — high-FODMAP ingredients for IBS), or good match. Hard-block flags always come from verified, human-reviewed data; an unverified AI-proposed tag can never set one.

A worked example: a wheat-based snack bar

With a Celiac profile active, the same bar produces three different flags depending on what a given ingredient tag says.