Nobody decided your wearable isn't a medical device

The regulation names software, and it is broader than you think

Regulation (EU) 2017/745 defines a medical device in Article 2, and software appears in the list of things that can be one, alongside instruments and implants. The trap is the list of purposes that follow. Most people assume it means treatment. It also covers *monitoring, prediction and prognosis* of a disease — and monitoring is precisely the word a wearable company would use to describe its own product on a good day.

Nothing in that definition turns on how sophisticated the software is, what sensors it uses, or whether a clinician is involved. It turns on what the thing is intended to do.

And intended purpose is read from your marketing

This is the part that surprises people. Article 2 defines intended purpose as the use the manufacturer intends according to the data it supplies — on the label, in the instructions for use, and in its promotional or sales materials or statements.

Promotional materials. Sales materials. Statements. A landing page is all three. So is a retention email, an onboarding screen, a press quote, a founder's post. The line between a wellness app and a regulated device is not drawn by the engineering; it is drawn by a sentence somebody wrote to improve conversion, in a meeting with no engineer in the room, and the regulation reads that sentence as evidence.

Which is why the whole category writes the way it does. "Not intended to diagnose, treat, cure or prevent any disease" is not legal decoration. It is the manufacturer declining to state an intended purpose that would classify its own product.

Meanwhile the output does change what people do

It would be convenient to argue that none of this matters because nobody really listens to a ring. That is not what the clinical literature says. In 2017 Baron and colleagues published a case series naming orthosomnia: patients who arrived at sleep clinics distressed by their tracker data, convinced their sleep was broken because an app had said so — and sleeping worse because of the worry. Some were resisting clinical advice in favour of the number on their wrist.

That is a small case series, not a population estimate, and it should not be inflated into one. But it is published, peer-reviewed evidence that this output reaches the part of a person that makes decisions. Anyone who has watched someone skip a session because a score was low already knew. The literature simply means you cannot wave it away.

So: a product that influences behaviour, classified by its own adverts

Put the two halves together. The output demonstrably reaches behaviour. The classification is decided by the manufacturer's own promotional language. And there is no step in a normal product process where anybody checks whether the second has quietly caught up with the first.

That is the uncomfortable bit, and it is not a scandal — it is a structural gap. Nobody is being reckless. It is simply that the decision has no owner, and decisions without an owner get made by whoever writes the copy.

What this product refuses to do

We build one of these. So rather than offer an opinion about where the line should sit, here is where we have put ours, in the code rather than in a paragraph.

*When a device sets your heart rate, Vitra stops drawing conclusions from it.* Tell it you have a pacemaker, an ICD or beta-blockers and it keeps showing your HRV and resting rate — they are your numbers — but it stops reading them as nervous-system state or as fitness, and removes them from your scores. A number your body does not set is not a read on your recovery, and a confident verdict about it every morning is worse than none.

*It will withhold today's scores until you have rated your own morning.* Optional, off by default. The reason is that a score you read first colours how you answer afterwards, and an anchored self-rating cannot be compared to anything. It is the only way to find out whether the ring agrees with you, and it requires the app to shut up first.

*Every measurement carries a line saying what it is made of. The score pages deliberately carry none* — because a score is a summary of other numbers, not a measurement of anything, and explaining it would dignify it.

None of which is a legal argument

To be plain, because this is a post about statements and their consequences: Vitra is not a medical device. What it shows is not a diagnosis, not a prescription, and not a substitute for a clinician. If something about your health concerns you, ask one. The refusals above are product decisions about what software should be willing to assert — they are not a compliance strategy, and we are not pretending they classify anything.

The question in the title stands for the category rather than for us. If you make one of these things, the useful version of it is narrow: who, specifically, signed off on the sentence at the top of your pricing page — and did anyone ask them what it means?

Frequently asked questions