My aunt is eighty-three and lives on her own in a terraced house in Leicester that she has no intention of leaving. There is a family WhatsApp group, and a good proportion of what happens in it is her children and nieces quietly worrying about her while being careful not to let her see us doing it.
Last year one of my cousins bought her a sensor system. Little discs on the doors and the kettle and the medicine cupboard, an app on our phones, and a gentle promise that we'd know if something was wrong. It worked exactly as advertised. We could see when she'd got up, when she'd made tea, whether she'd opened the pill drawer.
She hated it. Not loudly — she's not a loud person — but within about six weeks the discs had been "knocked off by the cleaner," all of them, in an impressive coincidence.
I've thought about that a great deal since, because it taught me something I now see everywhere in this market. The product worked. It just wasn't for her.
Who's actually paying
Almost every piece of AI-adjacent technology sold into ageing is bought by the adult child and used by the parent. The buyer's problem is anxiety. The user's problem is entirely different — she isn't anxious, she's fine, and what she wants is for everyone to stop treating her like a project.
Once you see this you can't unsee it, and it explains why so much of the category feels slightly off. The features are optimised for the reassurance of the person holding the phone: dashboards, streaks, alerts, activity summaries. The person in the house gets no interface at all. She's not a user of that product. She's the data source.
Which is a very strange thing to do to an adult who ran a school office for thirty years and does the cryptic crossword faster than I do.
I'm not against remote monitoring — falls are genuinely how independence ends for a lot of people, and a system that gets someone found in twenty minutes instead of nine hours is not a gimmick. But there's an enormous difference between "this alerts someone if I fall" and "this reports my day to my children," and the products blur it deliberately, because the blur is what sells.
What she does use
Here's the interesting part. She's not a technophobe. She uses things constantly — they're just the ones where she's the one being helped.
She photographs letters and has them read out to her, because the print on the pension correspondence is absurd and her eyes are tired by four o'clock. She dictates messages rather than typing them, at length, with punctuation spoken aloud in a way I find delightful. After a hospital appointment last winter she recorded the consultation, with permission, and got a plain summary of it afterwards — which mattered because she'd taken in about a third of what the registrar said, as most people do in that chair.
None of that is marketed to older people. There's no "silver" branding on any of it. It's just ordinary technology that happens to remove a specific, concrete obstacle, and she found most of it herself.
The pattern seems to be: things that restore capability get adopted, things that transfer oversight to someone else get resisted. That's not a subtle distinction and yet the sector keeps building the second kind.
The companion question
I'll be honest that I don't know what I think about the companionship products, and I'd distrust anyone who's certain.
Loneliness in older people is a genuine health crisis, not a soft issue — the physiological effects are on par with things we take extremely seriously. A device that talks to someone every day, remembers what they said yesterday, asks about the grandchildren by name, is measurably better than the silence it replaces. I've read the research and I've spoken to people who work in care homes, and they're mostly not sentimental about it. If it helps, it helps.
What I can't get past is the direction it lets things drift. It's very easy to imagine a commissioning decision where a companion product is funded because it's cheaper than the visiting service it quietly displaces, and nobody ever writes down that trade because nobody ever proposes it out loud. That's how it would happen. Not a decision — a substitution, made on a spreadsheet.
So my position, for now, is: fine as an addition, never as a saving. And I'd want the person it's for to have been asked, in a real conversation, whether they want it.
The test I'd apply
If you're building in this space, one question. Can she turn it off?
Not "is there a setting." Can the eighty-three-year-old woman in the house decide, without a phone call to her family, without a fuss, without anyone being informed, that today she'd rather not be observed — and does the product still work for her afterwards?
If yes, you've built something for her. If no, you've built something for us, and you should at least be honest in the marketing about which one it is.
My aunt is fine, by the way. She's had a wrist alarm since the spring, which she chose herself, in the shop, having asked several sharp questions about who gets called first. That one's still on.
