Health Notice

Health Notice & Estimate Limitations

MeuMenu is an app by OXETECH TECNOLOGIA LTDA. This notice explains what the app's numbers and images are — and what they are not. It supplements the Terms of Use and the Privacy Policy.

1. Read this first

Every nutrition value MeuMenu shows — portion weight, calories, protein, carbohydrates, fat, fiber, and sodium — is an estimate generated by artificial intelligence from an image. It is not a measurement, not a lab analysis, and not medical or nutritional advice. Real margins of error exist and are part of how the app normally works. Do not use these estimates for clinical decisions. If you have diabetes, kidney disease, a food allergy, an eating disorder, or follow a prescribed diet, talk to a healthcare professional before relying on the app.

2. How the estimate is produced

• You photograph your plated meal. An AI model identifies what appears to be in the image, estimates the weight of each portion by visual comparison, and calculates nutrients from that estimated weight.
• You can enter the plate size or the portion weight yourself. That reduces one source of error, but the result is still an estimate. If you enter nothing, the AI estimates the weight — that is the default.
• Nobody weighs your food. There is no scale, no sample analysis, and the value is not a measurement of the dish in front of you.

Two photos of the same dish, taken from different angles, can produce different numbers — and neither is the true measure of the meal.

3. Where the estimate errs most

Estimating weight from a photo is inherently imprecise. The biggest sources of error:

• Portion weight — the largest of all. A photo is flat; food has height, and the model sees little of it. Because weight is the basis of the whole calculation, a weight error shifts every nutrient together, in the same direction.
• Cooking fat, sugar, and salt. Oil in the sauté, butter in the sauce, sugar in the glaze, and kitchen salt do not show up in the image. Sodium is one of the most fragile values for this reason.
• Sauces, broths, and creams. Appearance cannot tell thickened milk from butter-heavy cream.
• What is hidden. Lasagna layers, rice under the meat, fillings — the model only estimates what it can see.
• Fried and breaded foods. Absorbed oil varies with preparation, and the photo cannot answer that.
• Poor photo conditions. Low light, a low grazing angle, a half-eaten plate, delivery packaging: less information, more guesswork.
• Regional dishes and uncommon recipes. When the model does not recognize a dish, it answers with the closest thing it knows.

Read the value as an order of magnitude and a trend over time, never as the exact number for one meal.

4. Food allergies — never rely on the app

MeuMenu does not reliably detect allergens and must never be used to decide whether a dish is safe for someone with a food allergy. Ask the people who prepared the food. Every time. Including at places where you have eaten before.

• The ingredient list shown is inferred from the image or from the menu text. It does not come from the restaurant and nobody has verified it. It can omit an allergen that is present and include an ingredient that is not.
• Traces and cross-contamination never show up in a photo — the shared fryer, cutting board, and spoon are invisible.
• Preparation ingredients disappear into the dish: milk in the mash, egg in the pasta, tree nuts in the sauce, shellfish in the stock.
• The app does not know your allergy, does not block dishes, and does not mark anything as safe.

If you have an allergic reaction, seek emergency care immediately — call your local emergency number (911 in the US, 192 — SAMU — in Brazil).

5. Not medical or nutritional advice

MeuMenu is an informational tool. It does not diagnose, does not indicate treatment, does not prescribe diets, and does not replace a physician or a registered dietitian/nutritionist. In Brazil, prescribing diets is an activity legally reserved to licensed nutritionists (Law No. 8,234/1991). MeuMenu is not a medical device.

Do not base clinical decisions on the app's estimates. In particular:

• Carb counting for insulin dosing — a portion error becomes a dosing error. Use the methods your care team taught you.
• Renal diets — besides sodium being fragile, the app does not estimate potassium or phosphorus.
• Phenylketonuria (PKU) — the app does not calculate phenylalanine.
• Celiac disease — gluten hides in thickeners, stocks, and cross-contamination; the app does not classify dishes as gluten-free.
• Pregnancy with dietary restrictions — the app does not assess food safety (doneness, pasteurization, foods to avoid).
• Eating disorder treatment — see section 6.

6. Eating disorders

Calorie counting helps some people and harms others. If counting fills much of your day, brings guilt after meals, makes you avoid eating with other people, or leads to compensating afterward, pausing the app is a legitimate decision — and talking to someone helps more than adjusting the estimate.

• In the US: call or text 988 (Suicide & Crisis Lifeline, free, 24/7).
• In Brazil: call 188 (CVV, free, 24/7) or visit cvv.org.br; free public mental health care is available through CAPS/SUS.
• Your care team — the dietitian, physician, or therapist who already knows you — is best placed to say whether and how to use a tracking app.
• In an emergency, call your local emergency number (911 in the US, 192 in Brazil).

7. Dish images are AI-generated

When the app shows an image of a dish from a menu, that image is a synthetic illustration created by AI from the dish's name and description. It is labeled as AI-generated and carries the SynthID watermark embedded by Google.

• It is not a photo of the real dish and did not come from the restaurant.
• The dish that reaches your table may differ in portion size, presentation, and sides.
• Do not use the image to judge portion size or to identify ingredients.

8. Sensible use

• Follow the trend, not a single meal — thirty estimates show a direction; one carries all the error of that photo.
• Compare options rather than pinning down absolute values.
• Correct the portion when you know it — enter the plate size or weight, or adjust mentally: you were there, the app was not.
• Help the photo: good light, the whole plate in frame, before you start eating.

9. Wrong or inappropriate results

If an estimate seems absurd, the ingredient list looks wrong, or a generated image is inappropriate, write to suporte@oxetech.cloud with the dish name, what the app showed, and — if you still have it — the photo you used. A health emergency comes first; the report can wait.

Responsible for the app: OXETECH TECNOLOGIA LTDA, Brazilian company registry (CNPJ) 67.217.613/0001-56, Av. Miguel Castro, 1264 — Nossa Senhora de Nazaré, Natal/RN, CEP 59.062-000, Brasil. Contact: suporte@oxetech.cloud. Data Protection Officer: OXETECH TECNOLOGIA LTDA.

10. Effective date

• Version: 2026-08-31.1, effective August 31, 2026.
• When this notice changes, the new version is published here with an updated version number and date; material changes are also highlighted in the app.
• To learn what happens to your photos and data — including that photos are processed on our servers in Brazil and sent to Google for processing — see the Privacy Policy.