Retain / Save
Month-4 cliff companion coach (GLP-1)
The medication is working — it's the member who goes quiet around month four, and a companion coach is what carries them to the end of the course.
What it does
GLP-1 programs see the same cliff: members start strong, results plateau, the novelty fades, and around month four they drift — first quieter, then gone — even though the clinical value sits in completing the full twelve-month course. The medicine itself is commoditized; what retains is the coaching relationship. The agent acts as a companion coach across the cliff: preemptive encouragement when the plateau approaches, a personal "noticed you've been quieter" reach-out when engagement decays, and adherence nudges that keep doses on schedule — so the program keeps the member it already paid to acquire.
How it works
- 1Trigger. the member's program month approaches the cliff, their engagement decays below their own baseline (fewer logs, fewer opens, quieter chat), or dose logging slips.
- 2Decision. the agent draws on the member's memory — their goals, past wins, how they like to be spoken to — and picks the right companion move: a plateau reframe before the frustration peaks, a quiet-user check-in, or an adherence nudge. A judge gates every unprompted message for tone and safety.
- 3Action. the touch arrives as a push or in-app chat message that reads like a coach who remembers them, not a campaign — and opens a real conversation inside the app.
- 4Follow-through. the conversation continues with the in-app coach; wins and concerns are written back to memory for the next touch; clinical concerns escalate to a human; retained months are measured against a holdout so the program sees the lift the coach actually created.
Configuration
How the agent is wired for this use case.
- Messaging channel · send the judged companion touch as push or in-app chat and continue the conversation
- Customer record · read goals, progress history, and program dates; flag clinical concerns
- Scheduling system · read program-month milestones and dose-logging schedule, detect dose-log gaps
- Memory store · read coach-conversation history; write wins and concerns back for the next touch
- Clinical escalation path · route clinical questions to a human
What you need
The inputs this use case runs on. Your channels stay yours; the agent supplies the judgment.
Signals
program-month milestones, engagement-decay detection against the member's own baseline, dose-log gaps, weight-trend plateau
Data
program start date and course length, goals and progress history, coach-conversation memory, notification consent
Guardrails
judge review on every unprompted message; clinical questions route to a human; companion tone, never pressure; frequency caps; suppression once the member re-engages; holdout measurement of retained months
Metrics it moves
- churndown, by defending the single steepest drop-off point in the program
- adherenceup, through dose nudges and plateau coaching that keep members on schedule
- dau-mauup, as quiet members are pulled back into the app conversation
- ltvup, with each member carried from the cliff toward the full course on high-CAC economics
Related use cases
Drop-off-window cadence coach (GLP-1)
the full map of drop-off windows, not just the steepest one
Adherence guardian for medication programs
the safety-first sibling for clinical risk signals
Commitment check-in
the same companion muscle for goals outside medication programs
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