Reading a photo: what AI can and cannot see in your pet

When you add a photo to a check, Petwell reads it with a vision-capable model, alongside your description. People are often surprised by how much a model can pick up, and it is just as important to be clear about what it cannot.

What a photo can show

A good, well-lit photo carries real signal. A model can often note redness, swelling, the look of a skin patch, discharge from an eye, or how a pet is holding a leg. Combined with your words, that helps place a symptom in context and shape the right questions for the vet.

• Surface things you can see: redness, swelling, a rash, a wound, discharge.
• Rough location and spread of a visible problem.
• Posture and how a pet is favoring a leg or holding itself.

What a photo cannot show

A photo is a flat, single moment. It cannot feel a temperature, listen to a heart or lungs, feel an abdomen, or run a blood test. It cannot see under the skin or inside a joint. Lighting, blur, and angle can all mislead. This is exactly why Petwell treats a photo as a starting point and prepares you for a vet, rather than pretending to be one.

A model can describe what a photo shows. Only a vet can examine your pet.

Why model-agnostic matters here

Vision models are improving quickly. Petwell routes across the best available models through a thin layer, so as they get better at reading images, the guidance gets better without you changing anything. We are not tied to one provider, and we are not tied to one moment in time.

The goal is not a clever demo. It is calm, honest guidance that helps you decide what to do next, and a vet visit that starts from a clear picture.