The honest reason I built Tailways.
I didn't set out to build a dog company. I set out to build wearables that work across species.
That sounds abstract, so let me be concrete. I'd been building SYLON, which senses a person's physiological state and gently nudges it toward calm or focus. Sense, understand, influence, learn. Once that loop existed and I could feel it working on myself, the obvious question was who else it could serve.
The answer was sitting in most of my friends' houses. Dogs are the animals closest to us. They live in our homes, and nearly everyone I know counts theirs as family. If the loop worked at all, it should work for them. So I started asking around, and I kept hearing the same thing: people who loved their dogs, could tell something was wrong, and had no way to understand it or help.
Then a friend asked me directly, about her own dog.
She'd leave for work in the morning and he'd fall apart. Not mildly unsettled. Genuinely distressed, for hours, every day, until she came back. She knew it was happening and she couldn't stop it, and she couldn't even tell me how bad it got on any particular day, because she wasn't there to see it. Nobody was.
That's the moment this stopped being an interesting engineering problem. It was heartbreaking, and it was completely invisible.
So we started where everyone starts. We tried to measure his heart rate.
We failed, and not narrowly. Fur defeats optical sensing in a way that's hard to appreciate until you've spent weeks on it. A usable signal means shaving the animal or strapping something tight enough to be miserable to wear all day, and the exact moments you most want a reading, a distressed dog pacing and panting, are the moments the signal is worst. We could have kept pushing. Instead we accepted the approach was wrong.
The problem hadn't gone anywhere though, so we changed the question. Instead of borrowing a human metric that doesn't survive contact with an animal, what would you measure if you designed for the dog from scratch?
What we landed on is an emotional spectrum. Not a number an owner has to interpret, but a state they already understand: sleeping, calm, alert, anxious, active. And measured against that particular dog's own baseline, because a greyhound's stillness is nothing like a terrier's. The useful question was never "what is the absolute reading". It's whether this dog has dropped below what's normal for him, and for how long.
We read it with 24GHz radar, which sees movement and micro-movement straight through the coat, down to the rise and fall of breathing, with nothing touching skin and nothing the dog can feel. Nobody has published how to interpret radar returns from a dog. There's no reference data to work from. That was the hard part, and it's most of the reason I wanted to do it.
I'll tell you what we haven't proven, because I'd rather you heard it from me. The hardware works: twenty-nine hours of continuous recording on a live dog, an eight-hour unbroken session, three interruptions in the entire record. But I can't yet tell you how accurate the classifier is, because that needs someone other than me watching a dog on video and scoring it blind against what the device said. That work is next, and when I have the number I'll publish it, whatever it says.
There was also a week where I was certain I'd had a hardware failure. I wrote it up, built an operating rule around it, then went back through the device's own telemetry and found the battery sensor had been reading backwards the whole time. My diagnosis was wrong. I'd rather show you that than a page of claims.
There's more we're working toward. Understanding what caused an episode rather than only detecting it. A model of each individual dog that sharpens the longer it's worn. Those aren't finished, and I'm not going to write about them as though they are.
Here's the part I didn't see coming.
I'd never had a dog. Not as a child, not as an adult. I came to this as an engineer with a loop that worked on humans and a question about where else it could go.
I spend time around dogs now. I can read them a little. I'm friends with a few of them, which is not a sentence I expected to write.
I still can't ask one how it's doing. That's the whole problem, and it hasn't changed. But I'm a lot closer to being able to tell.
Shared as personal experience, not a veterinary claim. Tailways is experimental and is not a medical or veterinary device.