How TreatCheck Food Safety Checking Works

How Food Safety Verdicts Are Generated

When you check a food in TreatCheck, the system sends the food name (or identified ingredients from a barcode or photograph), along with your pet's species, breed, age, weight, health conditions, and any noted allergies to an AI model. The model draws on its training data — which includes a broad range of veterinary and nutritional sources — to generate a structured safety assessment specific to your pet's profile.

The verdict consists of a safety level (Safe / Caution / Unsafe), a paw score from 1 to 5 reflecting overall nutritional suitability, a brief reasoning summary, a nutrient assessment, any specific warnings relevant to your pet's health conditions, and an appropriate treat size recommendation. For 'Unsafe' or 'Caution' foods, a toxicity severity indicator is shown.

What Safe, Caution, and Unsafe Mean

Safe: The food does not contain any known toxic compounds for your pet's species, and is considered generally appropriate in reasonable quantities. 'Safe' does not mean unlimited — it means no known safety concern.

Caution: The food has some characteristics that warrant moderation or awareness. This might mean it is high in fat (pancreatitis risk), high in sugar (obesity, dental), contains a compound that is safe in small amounts but harmful in large quantities, or is safe for most pets but has a specific concern given your pet's health conditions.

Unsafe: The food contains compounds known to be toxic to your pet's species, or carries a substantial risk given your pet's specific health profile. Unsafe foods should not be fed under normal circumstances.

Caching, Accuracy, and Limitations

TreatCheck caches food safety results to improve speed and reduce API costs — if the same food has been checked for the same species before, the stored result may be returned. This means results are consistent and fast, but it also means results may not reflect the most recently published veterinary guidance if it has changed since the result was last generated.

The most important limitation to understand is that TreatCheck's AI-generated results are not a replacement for veterinary advice. They represent a well-informed starting point, not a professional clinical assessment. Your individual pet may have specific health conditions, intolerances, or medication interactions that cannot be fully accounted for without direct veterinary knowledge of your animal. Always consult your vet if you have specific concerns or if your pet has a complex health history.

Reporting Concerns and Improving Results

If you believe a food safety result is incorrect or significantly out of date, you can report it using the feedback option on the result screen. Our team reviews reported results and updates them where appropriate. Community feedback helps improve the accuracy of the shared result cache over time.

For foods where you need the highest level of confidence — for example, before making a significant dietary change for a pet with kidney disease, diabetes, or heart disease — always follow up with your registered veterinarian. TreatCheck is designed to handle everyday food checking quickly and helpfully, not to replace the expert knowledge of a vet who knows your pet.