A patient voice, every time you ask.
Remio is a voice companion for people living with Alzheimer's and dementia. It answers the same question the tenth time as gently as the first.
The problem
Living with dementia means losing access to things you have known your whole life. Where the coffee is kept. Who the person on the phone is. What you are supposed to be doing this afternoon.
The answers exist — a family member knows all of them. But they cannot be present every hour of every day, and asking the same question repeatedly is exhausting for everyone involved. What gets lost is not just information. It is independence, and the dignity of not having to ask.
How it works
-
The family fills in a profile
Personal details, family members and how they are related, the home address and where things are kept, and step-by-step daily routines. This is set up once, by a relative or carer, and can be updated at any time.
-
The person asks out loud
No menus, no typing, no learning an interface — which matters, because learning a new interface is precisely what dementia takes away. The person simply asks, the way they would ask anyone in the room.
-
Remio answers aloud
In the person's own language, in short clear sentences, in a warm and unhurried tone. It never corrects, never points out that the question has been asked before, and never makes anyone feel they have failed.
Never inventing an answer
This is the design constraint everything else is built around. If a language model does not know where the coffee is kept, the tempting behaviour is to produce a plausible guess. For this audience that is not a harmless error — it sends a person with memory loss searching the house for something that was never there, and erodes the trust that makes the product worth having.
Remio answers only from the profile the family provided. When it does not know, it says so plainly and suggests contacting a named family member. That behaviour is enforced in the system's design, not left to chance.
Status
Remio is a working prototype. Profile management, spoken questions and spoken answers function end to end. It is not yet available to the public, and is not a medical device.
In development:
- Continuous speech recognition
- Recognising familiar voices
- Always-available background listening
Privacy and consent
A product that listens in someone's home carries real responsibility. Recording other people — visiting family, carers, neighbours — requires their knowledge and consent, and we treat that as a design requirement rather than a legal footnote. Explicit consent flows are being built before any always-listening capability ships.
Try the prototype
A browser version of the app is available to try. It needs an access key, which we provide on request — the demo talks to the same service as the real app, so access is limited. Voice input is not yet connected; questions are typed for now.
Working in dementia care?
We are looking to test Remio with families and care professionals. If you support someone living with dementia, or work in the field, your perspective would shape this far more than ours can alone.