All use cases

Retain / Save

Adherence guardian for medication programs

A patient's check-ins go quiet right as side-effects peak — the agent reaches out with the coaching touch that keeps them in the program.

Get aheadRisk triggerBehavioral triggerTime-based triggerIn-appPushChatHealth & TelehealthSubscriptionGLP-1 & Peptides

What it does

Medication programs lose patients in predictable risk windows, and most of those patients quit silently — they never message support, they just stop. In GLP-1 programs, more than half of patients discontinue within the first year, usually during dose escalation when side-effects feel worst and results have not arrived yet. The agent watches the program's own signals — titration stage, logged side-effects, weight plateaus, check-in cadence, engagement decay — and reaches out with a safe, encouraging coaching touch before the quiet patient becomes a cancelled one. Anything clinically serious goes to a clinician, not a chat reply.

How it works

  1. 1
    Trigger. server-side risk signals fire — a side-effect logged during a dose increase, check-ins going quiet, a weight plateau, engagement decaying across the app.
  2. 2
    Decision. the agent weighs the signal against the patient's memory (program stage, what has already been said, prior outreach and how it landed) and clinical safety policy; every unprompted message is judge-gated for safety and tone before it sends.
  3. 3
    Action. a coaching message in-app, by push, or in chat that normalizes what the patient is feeling and offers what actually helps — practical and encouraging, never clinical advice.
  4. 4
    Follow-through. red flags escalate to a clinician with context attached; the patient's reply is written back to memory; if the patient re-engages on their own first, the stale outreach is cancelled. Impact is measured against a holdout group, so the program sees what the guardian actually saved.
Example
AgentHi — nausea during a dose increase is one of the most common things patients report at this stage, and for most people it settles. Smaller meals and staying hydrated tend to help. How are you feeling today?
CustomerHonestly, I've been thinking about stopping. It's been rough.
AgentThat feeling is really common at exactly this point, and it's usually temporary — patients who get through this stretch tend to see the side-effects ease. If it doesn't let up, your clinician can look at adjusting the plan. Want me to set that up for you today?

Configuration

How the agent is wired for this use case.

Triggera server-side risk signal fires from the patient backend — a side-effect logged during a dose increase, check-ins going quiet, a weight plateau, or engagement decaying across the app.
Tools & actions
  • Patient backend · receives the risk signal and reads program stage, medication and dose history
  • Patient memory store · reads prior coaching conversations and outreach, writes the patient's reply back
  • Messaging channel (in-app, push, chat) · sends the coaching touch that normalizes the experience and offers what helps
  • Clinician escalation queue · routes red-flag conversations to a clinician with context attached
Autonomycoaching outreach runs unattended, judge-gated for safety and tone on every unprompted message, under a coaching-not-medical-advice clinical-safety policy; clinically serious signals are never answered by chat — they route to a clinician.
Channelsin-app · push · chat
Escalationred flags escalate to a clinician with context attached; if the patient re-engages on their own first, the stale outreach is cancelled.

What you need

The inputs this use case runs on. Your channels stay yours; the agent supplies the judgment.

Signals

titration/dose-stage events, side-effect logs, check-in cadence, weight or outcome tracking, engagement-decay events

Data

patient program stage, medication and dose history, consent state, memory of prior coaching conversations and outreach

Guardrails

clinical red-flag detection with clinician escalation on every conversation; judge gating on every unprompted message; coaching-not-medical-advice policy; frequency caps; holdout measurement so claimed saves are proven, not assumed

Metrics it moves

  • adherenceup, as patients caught inside the risk window stay on protocol instead of quietly stopping
  • churndown, because program cancellations concentrate in exactly the windows the guardian covers
  • ticket-deflectionup, since the proactive touch answers the side-effect question before it becomes a support contact, measured against the holdout
  • ltvup, as every retained month of a medication program compounds on already-spent acquisition cost

Related use cases

Retain / Save

Drop-off-window cadence coach (GLP-1)

scheduled touchpoints on the mapped drop-off calendar, versus this card's live risk signals

Prevent / Care

Clinical red-flag detection and safe escalation

the safety rail every guardian message leans on

Retain / Save

Missed-refill and replenishment outreach

the refill-event sibling of the same retention moment

See it on your own customer journey

Bring one drop-off, one churn cliff, or one silent segment. We will show you what a proactive agent with memory and judgment does with it.

Book a demo