All use cases

Prevent / Care

GLP-1 side-effect education with red-flag escalation

A patient writes "I've been so nauseous since my shot" at 11pm — the agent answers with real self-care, asks the questions that catch the dangerous cases, and pulls in a clinician the moment it isn't routine.

Close the loopGet aheadNext best thingBehavioral triggerRisk triggerWhatsAppChatIn-appPushSMSGLP-1 & PeptidesHealth & Telehealth

What it does

Most GLP-1 side-effects are mild and manageable, but patients can't tell mild from dangerous on their own, and they mention symptoms in passing — a chat, a WhatsApp reply — not in a triage form. Send every "I feel sick" to a clinician and the medical team drowns; answer with a generic bot and you miss the one report that was an emergency, or worse, reassure someone you should have escalated. The agent closes that gap: it acknowledges the symptom, shares the program's own knowledge-base self-care for the mild path, and asks structured severity and red-flag questions before it says anything reassuring. When the answers point to severe, unexpected, or out-of-pattern, it stops advising, drafts a clean summary, and routes to the medical team — then tells the patient the moment a clinician replies.

How it works

  1. 1
    Trigger. a side-effect mention in any channel — nausea most often — or a symptom pattern surfacing across the patient's logged history.
  2. 2
    Decision. the agent classifies severity against the program's clinical policy and the patient's medication stage, asking the structured red-flag questions first; conservative by design, it never reassures until the dangerous branches are ruled out.
  3. 3
    Action. for the mild path it shares knowledge-base self-care grounded in the program's content; for severe, persistent, or unexpected reports it withholds advice, drafts a summary, and files it to the medical team.
  4. 4
    Follow-through. the escalation carries the full history so the patient never repeats themselves, automated replies stand down on that thread, and the agent notifies the patient as soon as the clinician responds — closing the loop in memory.
Example
PatientFelt really queasy and a bit dizzy since my injection yesterday, is that normal?
AgentSome queasiness in the early days is common, and there's plenty we can do about it. Before I share anything though — has the dizziness come with chest pain, fainting, or trouble keeping fluids down even in sips?
PatientI couldn't keep water down this morning and feel lightheaded standing up.
AgentThank you for telling me — that combination is one I want your care team to see rather than handle here. I've sent them everything you've described and they'll reach out shortly. In the meantime, small sips and stay seated; if it worsens, message me and I'll escalate again straight away.

Configuration

How the agent is wired for this use case.

Triggera side-effect or symptom mention in any conversation, or a recurring symptom pattern detected across the patient's logged history
Tools & actions
  • Messaging channel · acknowledge the symptom, ask structured severity and red-flag questions, deliver the mild-path reply
  • Knowledge base · retrieve program-approved self-care content for mild cases, grounded with citation
  • Memory store · log the symptom against the patient's medication stage and prior reports, write the clinician outcome back
  • Clinical system · file a drafted, sourced summary into the medical team's queue and route severe or unexpected cases
  • Messaging channel · notify the patient when the clinician replies and reopen the thread
Autonomyacknowledgement, severity questioning, and the mild-path self-care reply run unattended behind a judge that gates every message for tone and safety; the policy is detect then escalate — the agent gives no dose advice, makes no diagnosis, and defaults to escalation whenever the severity is uncertain
Channelswhatsapp · chat · in-app · push · sms
Escalationany severe, persistent, unexpected, or red-flag report hands to the medical team with a drafted summary; automated replies stand down until a clinician owns the thread

What you need

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

Signals

in-conversation symptom and side-effect mentions, dose and program-stage events, the patient's message history

Data

medication and titration stage, prior symptom reports, persistent patient memory across coaching, support, and medical threads, consent state

Guardrails

clinical triage policy with conservative defaults; no dose advice or diagnosis in any automated reply; judge gating on every message; automated replies stand down once a clinician owns the thread; full audit trail of each severity decision and escalation

Metrics it moves

  • safe-escalation-rateup, as dangerous reports reliably reach a clinician instead of getting a reassuring bot reply
  • ticket-deflectionup, because the large share of mild, self-manageable side-effects is resolved with grounded self-care, not a clinician's time
  • csatup, since patients get an immediate, watched-over answer at the hour they actually ask
  • adherenceup, as confidently-managed early side-effects stop becoming silent quits

Related use cases

Prevent / Care

Clinical red-flag detection and safe escalation

the general detection rail this GLP-1 education layer wraps around

Prevent / Care

Nausea and GI comfort coach for the first weeks

the onset-window specialization for the most common symptom

Prevent / Care

Symptom-triage urgency-ranking layer

the layer above, ordering many symptoms by acuity

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