Open Food Facts first, USDA as fallback
Food Focus first checks Open Food Facts for a packaged-product barcode. If needed, it falls back to USDA FoodData Central. Missing tracked nutrient fields may then be AI-estimated per 100 g before save.
Sources & methodology
Food Focus estimates nutrition. It does not measure it.
Food Focus combines packaged-food database records with AI-assisted estimates. A barcode result can use both: database data for the product and available nutrients, then AI estimates only for tracked fields the record is missing. Results remain editable because every source can be incomplete.
Short answer
Where each result comes from
AI-assisted nutrition estimates
These inputs produce estimates from the information available. The app cannot see every ingredient, preparation method, or exact portion, so users can edit the result.
A heuristic, not a clinical score
Photo, gallery, text, and voice analyses can receive an AI-estimated Health Index intended as directional context. A barcode entry currently starts with a default zero rather than the same AI assessment. Neither is a measurement, diagnosis, laboratory result, or universal nutrition grade.
The app's Health Index trend can include those barcode default-zero entries in its chart calculation. Do not interpret a dip caused by a barcode zero as an AI judgment about meal quality.
A 7-day inference after 10 meals
After at least 10 meals are logged, Food Focus can infer possible nutrient gaps from the previous 7 days. These signals are not measured deficiencies and do not diagnose a condition.
Packaged food
How barcode lookup works
A barcode identifies a database record. It does not verify the amount eaten or guarantee that a record matches the current label in every region.
- 1
First source: Open Food Facts
Food Focus checks the Open Food Facts product database first. Open Food Facts describes its data as openly reusable and user-contributed, and warns that records are not guaranteed to be accurate, complete, or reliable.
Open the official Open Food Facts API documentation - 2
Fallback source: USDA FoodData Central
When the first lookup does not provide the needed result, Food Focus uses USDA FoodData Central as its fallback source for food and nutrient data.
Visit the official USDA FoodData Central website - 3
AI enrichment only for missing fields
If calories, protein, fat, carbohydrates, fiber, added sugar, or saturated fat are absent from the selected record, Food Focus may estimate the missing per-100-gram values with AI. Those fields are estimates, not database or label facts.
- 4
Gram amount and user review
A scan starts at 100 g or the last gram amount used for that barcode on the device. The product, package label, generated missing fields, and total grams eaten still need review.
Tracked values
What Food Focus tracks numerically
The key meal fields tracked as numbers are limited to the seven values below. A Health Index or nutrient-gap message is a separate interpretive signal, not another directly measured nutrient value.
- Calories
- Protein
- Fat
- Carbohydrates
- Fiber
- Added sugar
- Saturated fat
AI methodology
Guidance uses broad nutrition principles, with clear limits.
WHO reference
Food Focus links to World Health Organization healthy-diet guidance as a general reference. The material emphasizes broad principles such as adequacy, balance, moderation, and diversity. Food Focus has not published a versioned rule-by-rule mapping, and this reference is not WHO certification, endorsement, or individualized medical advice.
Read the official WHO healthy diet fact sheet
Food Memory
Food Memory is enabled by default on each device, and its Settings toggle is stored locally rather than synced account-wide. In the normal single-meal flow, it replaces the current gram value for each recognized, normalized ingredient rather than averaging older portions. The account keeps up to 200 recently updated entries, which can be renamed, edited, deleted, or cleared. Siri can save and update memory without pre-save review, while a multi-meal paste can sum duplicate normalized ingredients.
Why estimates remain editable
Images do not reliably reveal hidden oil, sauces, ingredient brands, cooking losses, or exact weight. Voice and text depend on the detail supplied. Barcode records describe a product, not the amount consumed, and missing barcode fields may be AI-estimated. Food Focus therefore lets users correct foods, portions, and nutrition values before relying on a log.
Read before using nutrition guidance
Food Focus is informational and is not medical advice.
Calorie and nutrient values may differ from the food actually eaten. The Health Index is heuristic, and a 7-day nutrient-gap inference cannot establish a deficiency.
Do not use Food Focus to diagnose, treat, or manage a medical condition. For medical or individualized nutrition decisions, consult a qualified healthcare professional.