Design exploration

Companion Health

Quiet care signals for the beings who can't speak for themselves.

The problem

Pet wellness technology answers the wrong question. AirTag tells you where your pet is. Vet apps store records you check once a year. Neither gives you a continuous, low-friction signal for how your pet is actually doing — and neither is built around consent or minimum necessary disclosure. The data that exists is either location or clinical. Nothing in between attends to the everyday relationship.

The anchor — PetTag

PetTag is the first holdable piece of Companion Health: an AirTag-class companion device with a low-power e-ink face. It shows a pet's name, a basic health signal — normal or needs attention — and an emergency owner contact. Nothing more.

The contact surfaces only when the device has been separated from its owner for an unusual amount of time. Not on demand, not by default. The design question that shaped it: what's the minimum a stranger would need to know if they found your pet? Answer that honestly and you've defined the privacy model for the whole platform.

Four disclosure states, in order of escalation: name only — name with health signal — name with health and contact — full emergency mode. Each state requires a different condition to activate. The device shows the least it can, until it must show more.

The platform

PetTag is the endpoint. Companion Health is the platform it belongs to: a system that gives you an ambient sense of how the animals and caregiving relationships in your life are doing — not where they are, not a feed of data to check, but a quiet signal that surfaces only when something matters.

The difference between surveillance and care is frequency and intent. A GPS tracker answers a question you ask constantly. Companion Health answers a question you shouldn't have to ask: it tells you when the answer has changed.

The consent model

Companion Health shares its philosophy with Attune: minimum viable disclosure, always with explicit conditions. Where Attune applies this to the human side of the relationship — tone controls, consent responses, emotional bandwidth — Companion Health applies it to beings who can't advocate for their own privacy. That obligation makes the design problem sharper, not easier.

Horizon — service animals

The gap is most acute for service animal handlers. A blind handler whose guide dog has wandered, a person with epilepsy whose medical alert dog is out of range, a deaf handler who needs non-audio separation signals — current technology fails all of them in ways that matter. Companion Health's consent-first, event-driven model points directly at this territory, though the platform is not yet built.

What I directed

I defined the disclosure model — the four-state escalation logic, the separation detection heuristic, and the consent framework that connects PetTag to the broader platform. This is a design exploration, not a working prototype.

What I learned

The most interesting constraint was the obligation. Attune is consent design for people who can express their own preferences. Companion Health is consent design on behalf of beings who cannot. That asymmetry is where the hardest and most interesting design problems live.