SCAN & DECODE

Scan a barcode. See the health story behind a product.

Food and cosmetics, scored 0–100 and colour-coded green, amber or red. We show exactly where the points went, what the label is saying, and what it means for you.

Food · CosmeticsSeconds to a scoreExplainable deductions

Try the score demo →Join the beta

Interactive demo

See how a score is built

Enter a sample barcode or pick a product. Every deduction and every label note is calculated live from the rules on this page.

POINT AT THE BARCODE

Try:

This web demo simulates a scan with manual barcode entry; camera scanning is available in the app. It uses fictional products and simplified scoring rules to explain the product logic. It is not nutritional, medical or safety advice. Real products require barcode matching, label verification and versioned rules.

…

Loading the demo

Enable JavaScript to see the full score breakdown.

Three steps

From barcode to explanation

Recognition, matching and explanation are separate steps. If any step fails, we never fill the gap with a guess.

Scan

The native camera reads EAN-13, UPC-A and GS1 barcodes on-device, so a weak connection still gets you the code.

Match & retrieve

Once matched, the score comes from a precomputed record. Common products return in a few hundred milliseconds, with no live LLM scoring.

Explain & advise

Break down deductions, translate the ingredient list, flag risks for specific groups, and suggest alternatives that weigh health and taste together.

Two models

Food and cosmetics need different rulers

Food is about intake and nutrient density. Cosmetics are about formulation and contact risk. We model and explain them separately.

Food: what you ingest

Starting from per-100g values and your actual portion, we score sugar, sodium, saturated fat, energy and NOVA processing, then credit fibre, protein and whole-food content.

Added sugarSodiumSaturated fatEnergy densityNOVAAdditivesFibreProtein

Output: 0–100 health score + every deduction + portion guidance

Cosmetics: what stays on your skin

We read the INCI list for high-concern ingredients, harsh surfactants, preservatives and fragrance allergens, then factor in formula complexity to produce an ingredient concern score.

Endocrine concernsHarsh surfactantsFragrance allergensPreservativesAlcoholFormula complexitySensitive-skin notes

Output: 0–100 ingredient concern score + ingredient flags + usage notes

The scale

0–100, green / amber / red

Colour expresses how strongly we would prioritise a product. It is not a binary “safe or toxic” verdict.

75–100

Green · Generally better

Most indicators sit in a reasonable range and there are few deductions. Portion and personal context still matter.

50–74

Amber · Worth noting

One or two clear deductions, or ingredients worth noting. Fine to choose, but not as a daily staple.

0–49

Red · Choose less often

Several indicators are high at once, or the formula contains many ingredients of concern. Look for an alternative.

Fruit cereal

SAMPLE LABEL · PER 100 G

Energy	430 kcal
Sugar	28 g
Sodium	320 mg
Saturated fat	6 g
Fibre	4 g

Ingredients: oats, sugar, wheat flour, vegetable oil, high-fructose corn syrup, flavouring, antioxidant, salt

Fictional sample for illustration only

Beyond the number: the label in plain language

Most people do not read labels because labels were not written for them. Ingredient lists, nutrition tables and INCI names are built for regulators and formulators.

01Ingredients are ordered by weight. What comes first is what there is most of. “Fruit flavour” may mean flavouring, not fruit.
02Sugar has many names. Sugar, high-fructose corn syrup, fruit concentrate and maltodextrin all count as free sugars.
03Turn “per 100 g” into your portion. Compare it with the daily reference intake to see what one serving really contributes.
04Read the INCI list. We translate names like sodium laureth sulfate and phenoxyethanol into cleansing power, preservatives and allergen risk.

Data & limits

What we can claim — and what we will not

Trust is the hard part of a scanning product. We would rather admit we do not know than show a confident wrong number.

Data sourcesGS1 barcodes, brand packaging, public food composition data and user contributions. Every record keeps its source and update date.

Reproducible scoringRules are versioned. Every point added or removed traces back to a specific metric, threshold and rationale.

No guessing when unmatchedIf a product is missing, we say so and invite a label photo for review instead of inventing a score.

Privacy firstBarcode and image recognition happen on-device where possible. Uploads require separate consent.

Not a substitute for adviceA score is a consumer reference, not a diagnosis, treatment or final safety verdict. Consult a professional for medical, allergy or pregnancy needs.

Category by categoryWe start with high-frequency foods and basic skincare, then extend to long-tail categories. Thin coverage is labelled as such.

FAQ

About Scan & Score

What happens if a product is not found?

If the barcode is recognised but the product is not in our database, we do not guess. We show that it is not yet covered and invite a photo of the ingredient list and nutrition table for review.

Is the score an official verdict?

No. It is a transparent, reproducible consumer reference, not a medical diagnosis or a substitute for a doctor, dietitian or regulator. Nutrition and toxicology evidence changes, so we publish the rule version and update date.

Can cosmetics have a health score?

We call it an ingredient concern score. It reflects high-concern ingredients, allergens and irritation potential. It is not a claim of medical benefit and not a final safety verdict. Sensitive skin should still patch-test.

Does it work offline?

Barcode recognition can run on-device. Product matching and scoring need a connection, while previously cached products remain available offline.

Will it recommend alternatives?

Yes. Alternatives weigh both health and taste preferences, with the reasons stated. Health is never the only ranking signal.

How broad is coverage in China?

Coverage expands category by category, starting with high-frequency SKUs. Long-tail products are added through brand data and user contributions. Categories with thin coverage are labelled clearly rather than hidden behind a single score.

Turn a scan into an explainable health decision

Start with one product: see what it means for you, then decide whether to buy it, how much, and whether there is a better alternative.

Book a demo Explore the Decision Agent