A quiet flat that looks after Margaret.
No cameras. No microphones-on-by-default. Nine small sensors learn Margaret's day — when she usually gets up, makes tea, settles into the lounge — and only ask for a human when something falls outside her pattern. Use the speed control to fast-forward to the 14:47 fall scenario, or step through the timeline minute by minute.
Loading the home…
Privacy-first by design
Movement, posture and routine — never images or audio without an explicit two-way call. Every sensor is consented, named, and shown to the service user and family on the same dashboard.
Patterns, not paranoia
The system learns Margaret's routine over the first 7–10 days, so a 03:00 trip to the bathroom is normal and a 14:47 hallway impact is not. Fewer false alarms means alerts are taken seriously.
Escalates with context
When a fall fires, the watch, AI Companion and family app all light up at once — each with the right detail for that audience. No-one is left chasing a ringing phone.
What's actually in Margaret's flat
PIR motion (×4)
Ceiling-mounted in bedroom, hallway, kitchen and lounge. Detect occupancy and broad activity.
Bed pressure mat
Knows when she's in bed and when she's up. Long absences trigger a check-in.
Hallway fall mat
Combines with the wearable's IMU to confirm a fall vs a dropped item.
Door sensor
Front door open/close — also doubles as carer NFC check-in/out.
Bathroom motion
Long-stay alerts (no movement for >30 min in bathroom) prompt a wellness check.
Panic pendant
Worn around the neck. One press routes through the watch and AI Companion immediately.