GitHub Surf VS PetSignal AI

PetSignal AI's answer:

Most pet apps ask you to describe the behavior in words. PetSignal reads the photo itself — facial tension, ear set, eye state, mouth shape, tail carriage, posture and surrounding context — and tells you what those cues usually mean.

Three things set it apart:

1. Separate dog and cat engines. Feline cues like tail flicks, slow blinks and ear rotation are read on their own terms, not force-fitted into a dog model.
2. Pre-bite warnings are stated plainly. Whale eye, freezing and lip curl get flagged explicitly instead of being softened into "he's just being shy" — those are the signals that precede bites.
3. Conservative by design. When a photo doesn't support a confident read, the report says so rather than inventing certainty. The methodology is grounded in published ethology research, including Turid Rugaas's work on calming signals.

PetSignal is an educational screening tool, not a veterinary diagnosis — when the signals warrant it, the report tells you to monitor closely or consult a certified behaviorist or veterinarian.

PetSignal AI's answer:

Because it works from what you already have — one ordinary phone photo — and it treats cats as a first-class species.

Most tools in this space either (a) only handle dogs, (b) ask you to type out a description of the behavior, which means the answer is only as good as your own reading, or (c) return a cheerful one-word verdict with no reasoning.

PetSignal returns the cues it actually detected, the reasoning that connects those cues to a behavior read, a confidence score it is willing to keep low, and one concrete next step. Pre-bite warning signals are named outright rather than softened. The methodology is anchored in published ethology research instead of vibes.

Practical differences: no app install, works in any browser, 30 free credits on sign-up (about 3 photo analyses) with no card required, one-time credit packs from $2.99 with no forced subscription, and a PDF export you can hand to a trainer or vet.

PetSignal AI's answer:

Everyday dog and cat owners — not professionals.

The core user is someone who noticed something and can't tell whether it matters: a new owner learning their pet's baseline, an adopter reading an unfamiliar rescue dog, an owner whose cat started hiding or flicking its tail, someone whose dog freezes around visitors.

A second important group is parents. "Is my dog okay around my toddler?" is a question people are afraid to get wrong, and the early warning signals are exactly the ones adults tend to misread as affectionate.

A third is the pre-vet moment: people deciding whether a behavior is worth a phone call to a professional. PetSignal is built to make that call easier to make correctly, not to replace it.

PetSignal AI's answer:

Dogs and cats warn us long before anything goes wrong — and people miss it.

Almost every bite report contains a few seconds of visible warning that nobody read: a lip lick, a head turn, a hard stare, a frozen body, a tail gone stiff. These are not subtle to an ethologist. They are nearly invisible to the average owner, who has been taught that a dog showing teeth is "smiling" and a cat flicking its tail is "playful."

PetSignal exists to close that gap. The cues are already in the photos people take every day — the reading, not the capturing, is the missing part. The engine was built around published body-language research, including Turid Rugaas's calming signals, with a deliberate bias toward caution: it would rather say "not confident, watch closely" than hand out a reassuring answer it cannot support.

It is a tool for noticing, not a substitute for a veterinarian or a certified behaviorist.

PetSignal AI's answer:

A modern web stack, deliberately browser-first so there is nothing to install:

• Frontend: Next.js (App Router) with TypeScript and Tailwind CSS
• Backend & data: Postgres via Supabase, with row-level security on user data
• Hosting & edge: Vercel, with Cloudflare for storage and edge services
• Analysis: a multimodal vision pipeline with species-specific prompting for dogs and cats, plus a scoring layer that keeps confidence deliberately conservative
• Payments: one-time credit packs, no forced subscription
• Reports: server-side PDF generation for exportable behavior reports