All use cases

Retain / Save

Early plateau pre-empt normalization

Around the two-month mark a patient quietly starts to wonder whether the program is working at all — and the agent gets there first, before that doubt hardens into a decision to stop.

Get aheadTime-based triggerBehavioral triggerPushIn-appChatGLP-1 & PeptidesHealth & TelehealthSubscription

What it does

Long before a GLP-1 patient sees a real weight stall, there is a softer moment — somewhere around the second month — where the first novelty has worn off and the thought "maybe this isn't working" begins to form. Most programs only notice once the patient has already gone quiet or the scale has already flattened, by which point the patient is half-decided. This agent does not wait for either signal. On reaching that early window it sends a brief, pre-emptive touch that names the feeling and normalizes it as a known part of the journey, so the patient never reaches the silent "nothing's happening" conclusion that drives the earliest churn. It is the first and gentlest point on the plateau timeline.

How it works

  1. 1
    Trigger. the patient reaches the early-plateau window (roughly the two-month mark) on their program calendar — a time signal that fires ahead of any weight stall or engagement drop, optionally tuned by light early-behavior context.
  2. 2
    Decision. the agent reads the patient's memory — start date, what they have logged so far, what the coach has already discussed, channel consent — and confirms they have not already had this reframe, so the touch lands once and on tone. A judge reviews every unprompted message before it sends.
  3. 3
    Action. a short, warm pre-empt lands on a program-owned channel (push, in-app, or chat) — naming the "is this working?" feeling and framing it as normal and expected this early, written to open the coach conversation rather than read like a campaign.
  4. 4
    Follow-through. replies continue inside the coach chat; any clinical concern in a reply escalates to the care team; if the patient has already re-engaged on their own or cancelled, the touch is suppressed; and the lift is measured against a holdout so the value is proven, not assumed.
Example
AgentYou're about two months in now — and right around here is where a lot of people start quietly wondering whether much is happening. For what it's worth, this stretch is a completely normal part of the journey, not a sign it's stalling. How are you feeling about it so far?
PatientHonestly yeah, I expected faster and I've been second-guessing it.
AgentThat's such a common place to be at this point, and it's not where the story ends — the early weeks rarely tell you the whole picture. Want to look together at what's actually shifted since you started, beyond just the scale?

Configuration

How the agent is wired for this use case.

Triggerthe patient reaches the early-plateau window (around the two-month mark) on the program calendar — a program_week_reached time event that fires ahead of any weight-stall or engagement-decay signal.
Tools & actions
  • Scheduling system · open the early-plateau touchpoint when the patient reaches the two-month window and queue the pre-emptive touch
  • Patient record / CRM · read program start date, early logging, conversation history, and channel consent
  • Memory store · check the patient has not already received this reframe; write the touch and outcome back for the next stage
  • Messaging channel · send the judged pre-empt on a program-owned channel and open the coach conversation
  • Clinical escalation path · route any clinical concern in the reply to the care team
Autonomywindow detection and the normalization touch run unattended, judge-gated on every unprompted message, using approved coaching content only under a never-new-medical-advice clinical policy; clinical concerns always route to a human, and sends are suppressed on cancellation or self-re-engagement.
Channelspush · in-app · chat
Escalationany clinical concern in a reply hands off to the care team; if the patient has already re-engaged or cancelled, the queued touch is suppressed.

What you need

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

Signals

program-week position (the early-plateau window), light early-behavior context, self-re-engagement and cancellation events.

Data

program start date, early-logging and conversation history, coach-chat memory, channel consent state.

Guardrails

judge review on every unprompted message; clinical concerns always escalate to a human; approved coaching content only, never new medical advice; reassuring tone, never pressure; once-per-window frequency cap; suppression on cancellation or self-re-engagement; holdout measurement of save-rate.

Metrics it moves

  • churndown, by reaching the patient before the early "nothing's happening" doubt turns into a quiet exit.
  • save-rateup, by normalizing the earliest plateau feeling at the first soft signal rather than after the scale or engagement has already dropped.
  • adherenceup, by keeping the patient in the conversation through the weeks where motivation first wobbles.

Related use cases

Retain / Save

Four-month drop-off re-engagement

the later window-based save that fires once engagement actually decays

Retain / Save

Weight-stall plateau intervention coach

the save fired by a real weight-stall signal rather than the calendar

Retain / Save

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

the full map of drop-off windows this earliest touch sits inside

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