How it works

From day 0 to day 30

The discharge itself starts the process, so nobody decides who gets followed up. Everyone does. What differs is how much attention each recovery turns out to need.

Day 1–3

What the patient is actually asked

The patient journey

That is the whole thing, about 30 seconds. A concerning answer can open one or two follow-up questions. A reassuring one ends the check-in there.

Day 1–3

How are you feeling compared with yesterday?

Worse

Thank you. Your care team has been notified.

Good morning, Anna

3 patients need your attention

  • Maria S.Needs attentionReported feeling worsenow

The patient journey

One example pathway. A hospital sets its own timing, messages, surveys, nurse tasks, waits, branches and exit rules for each pathway it approves.

  1. Day 0The discharge starts follow-up on its own. There is no enrollment step and nobody is selected.
  2. Day 1–3The first follow-up happens within 72 hours. An SMS, or a consented WhatsApp message, links to a short secure check-in in the patient's own language.
  3. Day 3–7Check-ins and pathway-specific education continue at whatever cadence the protocol sets.
  4. Week 2Each response updates where the patient sits. A recovery that is going well stays quiet.
  5. Week 3Problems that turn up are routed to whoever handles them, whether that is pharmacy, follow-up scheduling or a nurse call.
  6. Day 30The episode closes with a recorded outcome: what happened, what was done, and how it ended.

What the patient is actually asked

The patient taps a secure link from the message, which expires. It opens in the phone's browser, no app and no separate login. On one short page they answer:

  1. How are you feeling compared with yesterday?
  2. Any problems with your medications?
  3. Is something worrying you right now?

That is the whole thing, about 30 seconds. A concerning answer can open one or two follow-up questions. A reassuring one ends the check-in there.

Where a patient sits, rather than how likely they are to return

Every patient carries a state that updates as things change. It answers what needs to happen next, not how probable a readmission is:

  1. Stable
  2. Watch
  3. Needs attention
  4. Intervention
  5. Escalated
  6. Resolved

Responses, discharge information, medication issues, follow-up gaps and clinical rules move the state. A low-risk patient who couldn't collect their medication needs attention now. A high-risk patient with medicines in hand, clear red-flag instructions and a booked GP appointment can settle.

Every hospital starts with AI switched off, and only its own administrator can switch it on. Where it is on, a language model reads the discharge letter and the patient's replies in the background. Its reading is recorded beside every decision and never inside one: the follow-up is set by the ward's own risk call and by fixed, published rules, and the model cannot add contact, take it away, or move a patient up the list.

What happens to a problem once it is found

Each kind of problem has a route, an owner, and an outcome that has to be recorded before it closes:

Detected problemRouted to
Medication unavailablePharmacy / medication workflow
Medication confusionMedication reconciliation
Symptoms worseningClinical review
Follow-up not scheduledFollow-up workflow
Home care did not arriveCare coordination
Instructions not understoodTeach-back call
Transportation problemSocial support workflow
No response after configured attemptsHuman follow-up

The sequence is always the same: a problem is found, someone acts, it resolves, and the outcome is recorded. No episode simply stops.

Where the humans are

Book a demo