Living alone with a health routine means holding it all in your head. The tech offered for this is mostly a panic button or a camera — surveillance, when what people need is help.
What it would do
It sits still, listens when spoken to, and answers.
Medicine time. What to eat. The next doctor's appointment. It notices movement patterns and whether the medication was actually taken — and a reminder you can ask a question back to is a very different thing from an alarm.
Private by architecture
Movement data from inside a home describes when you woke, whether you ate, whether you fell. That class of data should be processed on the device — not streamed to a server.
That constraint is the research problem: useful inference on the compute an appliance can carry.
The resident decides
A device like this is usually bought by a worried son or daughter. The person it watches over is someone else.
So the resident holds the controls: what is sensed, what family sees, and the off switch. Consent is the first feature, not a settings page.
Where Lunas comes in
The reminders come from a real record: the medication list and appointments already in Lunas.
With the resident's consent, what the companion notices joins the same timeline a clinician can be shown — one health story, not another silo.
What this deliberately does not do
- Concept stage — nothing is built and there is no date
- Not a medical alert system and not a fall detector
- Reminders are not medication management; the regimen stays with a clinician
- The resident's consent comes before any caregiver's visibility