It watches the whole hospital around the clock, tells the right person what needs them, and prepares the next step as a card. It acts alone only on the low-risk chores your clinic allows.
Per signal and per role, with quiet hours and a daily cost cap.
1NotifyTell the right role, in the bell or on WhatsApp, with a one-line reason.
2PreparePut the action in an AI Inbox as a card: “Re-order insulin, 20 vials — Confirm”. A person decides.
3Auto-run, low risk onlyOnly what the clinic switched on: patient reminders, creating a task, escalating after a set number of hours.
Hard limits
What it will never do.
✕Write anything clinical: no prescriptions, orders, notes or results
✕Finalise a bill or move money
✕Message a patient who has not consented, or during quiet hours
✕Act beyond the level the clinic chose, per signal and per role
✕Spend past the clinic’s daily cost cap
An AI Inbox for every person
Each prepared step waits for the person whose job it is.
PreparedRe-order insulin, 20 vialsStock below par in the main storePrepared for the store keeperConfirmChange
PreparedReview 3 critical resultsWaiting more than 30 minutesPrepared for the doctor on dutyOpenSnooze
PreparedRemind 4 patients due tomorrowLow risk: runs by itself if the clinic allows itConfirmChange
Examples of planned cards, not real records.
Questions, answered
Is the 24/7 operator live?
Not yet. It is being built now, on the same assistant and confirm cards that are live today.
How is it different from Automations?
Automations, live today, sends the reminders and chores you switch on. The operator watches the whole hospital for signals, tells the right person and prepares the next step for them to confirm.
Can it act without a person?
Only for the low-risk actions a clinic enables — reminders, tasks, escalations — and never for anything clinical. Everything it does is logged with the reason.