Why We Built MyCanine360: Welfare-First AI Video Analysis for Dog Professionals

11 June 2026 · Rachel Trafford

Why We Built MyCanine360: Welfare-First AI Video Analysis for Dog Professionals

When we first saw what modern vision AI could do with video, the possibility was obvious. A model that can watch a training session, track a handler's mechanics, and read a dog's body language could save professionals hours and sharpen the feedback they give clients.

Then we looked harder at how those models actually form their interpretations, and we realised the possibility came with a serious catch.

The internet is not a behaviour textbook

A vision model does not learn dog behaviour from a peer-reviewed syllabus. It learns from whatever is available online, and the internet's view of dog training is a mixed bag, to say the least. Sound, science-led material sits alongside dominance myths, outdated advice, gimmicks and TikTok content, methods that no welfare-focused professional would endorse, and quite frankly many that are dangerous to humans and animals. The problem with intelligence gathering is that there is no filter, no parameters that aid progress in our field.

Left to interpret freely, an AI will reproduce that mixture with total confidence. It can label a stressed dog as "stubborn", read appeasement signals as defiance, or recommend an approach that an educated behaviourist would immediately reject. The analysis sounds authoritative. That is exactly the problem.

We were not willing to build a tool that gives confident, unsupervised behaviour advice drawn from the average of everything ever posted about dogs.

So we built it for professionals, on purpose

MyCanine360 is not a consumer toy that tells the general public what their dog is "thinking". It is built for educated dog training and behaviour professionals: the people who already hold the framework the AI lacks, and who understand the context of the environment.

That decision shapes everything. Our users can read a session in context, weigh what the model offers, and catch anything that does not hold up. The AI is there to accelerate and structure their work, not to replace their judgement.

Welfare guardrails and nuanced prompts

The interpretation a vision model produces depends heavily on how it is asked. So rather than letting the model interpret behaviour from scratch, we steer it.

We have built nuanced, layered prompts and ongoing refinements that push the analysis towards a welfare-first, science-led reading of what is on screen. 189 pages of prompts and climbing.

The guardrails are designed to keep the feedback grounded in modern learning theory and ethical practice, and to keep it away from the aversive interpretations that still circulate widely online.

This is the part of MyCanine360 that takes the most work, and the part we are proudest of. The vision capability is impressive on its own. The value sits in how carefully we frame what that capability is allowed to conclude.

The professional always stays in control

A machine should not have the final word on a living animal's welfare. That principle is built into the product.

Every analysis MyCanine360 produces is editable. The professional reviews what the AI has generated, adjusts it, corrects anything they disagree with, and signs it off. The branded, client-ready reports that leave the platform carry the professional's expertise, not an unchecked algorithm's opinion.

In practice that means the AI does the first pass and the heavy lifting, and the qualified human does the thinking that matters, the wingman concept. That is the relationship we think AI and behaviour work should have.

We are still learning, openly

We are honest about what these models can and cannot do. Vision AI is genuinely capable, and it is also imperfect, and the field is moving quickly.

So we treat MyCanine360 as something we are continually refining. We keep testing what the vision models can reliably see, where they fall short, and how to guide their interpretation within an ethical framework. As the models improve and as our understanding deepens, the guardrails and prompts improve with them.

That is a commitment rather than a finished claim. The technology will keep changing. Our responsibility to the dogs and to the professionals who care for them does not.

What this means for your practice

If you are a trainer or behaviourist, MyCanine360 gives you faster, more structured analysis of your sessions without asking you to trust a black box. You get a welfare-led first draft, full editing control, and reports you can confidently put your name to.

The AI handles the speed. You keep the standards.

If that is the way you already work, it is the way we built MyCanine360 to work too.