Scan a packet’s barcode and Ovessa shows how it fits a PCOS-friendly plate, as a letter from A to E, with the reasons spelled out. This guide explains where that letter comes from, what it can’t know, and why your allergies always come first.
Allergies first, always
Before any grade, Ovessa checks the product against the allergies you set in Settings (or during setup). The allergy screen outranks the grade; it doesn’t sit politely beside it.
- The 14 legally declarable allergens, such as peanuts, milk, gluten or sesame, are a hard rule: a declared match always reads as one to avoid.
- Your own free-text allergens get a weaker check against the declared ingredients, and the app says so plainly.
- Dietary preferences are advice only. A preference can never soften or hide an allergy warning.
The strongest reassurance Ovessa will ever give is that no allergens from your list are declared on this label. Never “allergen-free”, never “safe”. A database record shows what the label declares, not what is in the food, and formulas change. If you have an allergy, always read the physical label.
The grade is arithmetic, not AI
The grade is worked out on your iPhone from the product’s own label figures per 100 g (per 100 ml for drinks). It starts from a score of 100, subtracts points for each rule the product trips, and bands the result into five letters. Each threshold names its public source in the app:
- Sugars, saturated fat and salt: the UK front-of-pack traffic-light thresholds.
- Sugar in drinks: the upper band of the UK soft drinks levy.
- Refined carbohydrate: a carbohydrate-to-fibre ratio above 10:1.
- Trans fat and non-sugar sweeteners: WHO guidance.
The points each rule costs, the letter cut-offs and the cap for missing figures are Ovessa’s own calibration, not a published standard, and the app says so in its sources.
No model is involved and nothing is sent anywhere to grade a product. The barcode lookup that fetches the label is the only network request, and it carries only the barcode number.
What each letter says
The app keeps the wording calm. Food isn’t good or bad; it suits a PCOS-friendly plate more or less:
| Grade | What Ovessa says |
|---|---|
| A | A great fit for a PCOS-friendly plate. |
| B | Fits well on most days. |
| C | Okay today, in the amount you’d normally eat. |
| D | Worth going a bit easy on. |
| E | One to go easy on. |
Under the letter, each reason is one plain line, such as “High in sugar — 56.3 g per 100 g.”
When a label is missing figures
Ovessa won’t fill gaps with guesses:
- No sugar figure, no grade. “Can’t grade this one” is a real answer, not an error.
- Missing figures cap the grade. With one or two of the figures the rules need missing, the best possible grade is B; with three or more, C. The card says the grade is based on partial data, and the nutrition table shows a dash for each gap.
Where the numbers come from
- Barcodes look up the product in Open Food Facts, a public food database, with no cookies and nothing about you. Its records are shared under the Open Database License, and every product card shows the attribution. Products you’ve scanned before are saved on your iPhone and work offline.
- Meal photos and typed meals get calories and macros from the public USDA FoodData Central database. A model recognises what is on the plate and roughly how much, and helps pick the right USDA entry for each food; the numbers come from that entry. When USDA has no match, the model’s rough figure is used instead and marked “Approximate — not in the food database”, and the mark stays after a relaunch.
- Meal totals never invent a zero. A nutrient no source reports shows as “—”.
Photo portions are rough estimates, and you can correct every food and portion before anything is saved. Free includes 10 smart meals by photo or typed; Premium makes them unlimited. Barcode scans are always free.
Download on the App Store (opens the App Store in a new tab)