All use cases

Engage / Lifecycle

Psychological and identity support layer for GLP-1 patients

Rapid weight loss changes how a patient sees themselves and how others treat them, and almost no one is holding that conversation today.

Get aheadNext best thingBehavioral triggerRisk triggerChatIn-appPushGLP-1 & PeptidesHealth & Telehealth

What it does

The clinical side of a GLP-1 program is covered: the dose, the labs, the side effects. The psychological side is not. Patients lose weight fast, and their body image, their sense of identity, and the way friends and family react all shift faster than they can process — and when that goes unsupported, the patient quietly disengages from the program that is working. The agent adds a mood- and identity-aware support layer that recognizes these moments in everyday conversation, responds with grounded supportive coaching, and routes anything that crosses into clinical or mental-health territory to a human. It keeps the patient in a relationship through the part of the journey the protocol does not address.

How it works

  1. 1
    Trigger. the patient raises an identity or body-image theme in chat (feeling like a different person, discomfort with attention, complicated reactions from people around them), or a risk signal appears in their language or self-reported mood.
  2. 2
    Decision. the agent reads persistent memory — where the patient is in treatment, what they have shared before, how they like to be spoken to — and classifies the moment as supportive-coaching or escalate. A quality judge gates every unprompted message for tone, and a safety classifier scans for clinical or mental-health red flags.
  3. 3
    Action. for supportive moments, a short, grounded reply on chat or in-app that names the experience and normalizes it without minimizing it; for an unprompted touch, a gentle check-in when memory shows a recent identity-loaded event.
  4. 4
    Follow-through. the exchange is written back to memory so the next touch is continuous, not repetitive; any clinical or mental-health red flag halts the supportive flow and hands off to the care team, and a self-harm signal triggers an immediate stop, a helpline plus a human, and loop closure on the patient's safety.
Example
AgentYou mentioned it feels strange when people keep commenting on how you look. That reaction is really common on this journey — the change can land before you've caught up with it yourself. Want to talk through what's been sitting with you?
PatientYeah. Honestly I don't recognize myself in photos and it's messing with my head more than the weight ever did.
AgentThat's a real and valid thing to feel, and you're not alone in it. I'll stay in this with you. If it starts weighing heavier than you want to carry, I can also bring in someone from your care team who supports exactly this — no pressure, your call.

Configuration

How the agent is wired for this use case.

Triggeran identity / body-image theme detected in a patient message, or a mood / risk signal in self-reported check-ins, sourced from the messaging channel and the app backend.
Tools & actions
  • Messaging channel · sends the judged supportive reply and continues the conversation on chat, in-app, or push.
  • Memory store · reads treatment stage and prior conversation; writes each exchange back so support stays continuous.
  • App backend · reads self-reported mood and check-in history as trigger and context.
  • Knowledge base · retrieves grounded, non-clinical supportive framing for identity-shift and social-reaction themes.
  • Care-team escalation path · routes clinical or mental-health red flags to a human, with a crisis route for self-harm signals.
Autonomysupportive coaching runs unattended only after a quality judge clears tone; the agent never diagnoses, never counsels a clinical condition, and never advises on the medication — a clinical or mental-health red flag is a hard human handoff under a clinical-safety policy class, and a self-harm signal halts the flow for an immediate helpline-plus-human escalation.
Channelschat · in-app · push
Escalationany clinical or mental-health red flag hands off to the care team; a self-harm signal triggers an immediate stop, a helpline plus a human, and confirmed loop closure.

What you need

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

Signals

identity / body-image themes detected in messages, self-reported mood and check-in events, mental-health risk language.

Data

treatment stage, conversation memory, stated preferences for tone and contact, notification and messaging consent.

Guardrails

judge review on every unprompted message; clinical and mental-health red flags route to a human; self-harm crisis path with helpline and loop closure; supportive coaching only, never diagnosis, counseling of a condition, or medication advice; frequency caps and suppression once the patient re-engages.

Metrics it moves

  • churndown, by holding the patient through an identity-shift stretch that the clinical protocol leaves unsupported.
  • dau-mauup, as supportive moments open into real conversations rather than going unanswered.
  • adherenceup, a patient who feels held psychologically stays on the program that is working.
  • ltvup, a sustained support relationship carries high-CAC patients further into the course.

Related use cases

Engage / Lifecycle

Proactive health-coach companion

the always-on coaching backbone this support layer rides on

Prevent / Care

Crisis and red-flag detection for mental health

the crisis-escalation rail this layer hands off to

Prevent / Care

Over-restriction and disordered-eating guard

the sibling guard for an over-loss signal in the same audience

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