How we keep AI pet guidance safe

Petwell helps you understand what your pet needs, sooner, and walk into the vet prepared. It is guidance and vet-visit prep. It is not a diagnosis, and it is not a substitute for a veterinarian. Holding that line is not a caveat we bury at the bottom of a page. It shapes how the product is built.

What Petwell will never do

There are a few things Petwell will not do, by design, no matter how you ask. These are enforced in code and in the instructions the model runs under, so they hold even when a description is unusual.

• It will not name a diagnosis as fact. It speaks in terms of what a symptom could relate to.
• It will never give a medication name, a dose, or a treatment to administer at home.
• It always points medical decisions back to your vet, and it says so plainly.

How the emergency screen works

Some signs in a dog or cat should not wait: trouble breathing, collapse, suspected poisoning, severe bleeding, a seizure, a bloated hard belly with retching, or a cat straining and unable to urinate. Petwell screens for these two ways at once. The model is instructed to raise an emergency flag, and a separate check runs in plain code over what you wrote.

The code screen is negation-aware, so writing 'no trouble breathing' does not trip it. When either the model or the code sees a red flag, Petwell shows a clear note to contact an emergency vet now and does not bury it under other advice.

The most useful thing a guidance tool can do in an emergency is get out of the way and tell you to call for help.

Why honest limits build trust

It would be easy to sound more certain than the situation allows. We think that is exactly the wrong move for a pet's health. When there is little to go on, Petwell says so and keeps it short. It is calm, not alarming, and it is honest about what a description or a photo can and cannot tell you. That honesty is the point, and it is what makes the guidance worth trusting.