All use cases

Prevent / Care

Symptom-triage urgency-ranking layer

A patient lists three things at once — mild nausea, a sore injection site, and chest tightness — and the agent decides, in order, what they can handle at home and what a clinician needs to see first.

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

What it does

A GLP-1 patient rarely reports one clean symptom. They send a paragraph: some nausea, a lump where they injected, and one line buried in the middle that actually matters. As a program scales, every one of those paragraphs lands in the same clinical queue at the same priority, so the genuinely urgent case waits behind the harmless one and clinician time drains on questions a patient could have handled with guidance. The agent sits above the individual symptom flows as a triage layer: it separates what the patient describes into distinct concerns, ranks each by acuity, self-manages the low-acuity ones with approved guidance, and pushes the high-acuity one to a clinician first with the rest summarized underneath. It ranks urgency and routes; it never diagnoses or sets a dose.

How it works

  1. 1
    Trigger. a patient message (or a short logged history) describing one or more symptoms arrives on chat, in-app, or WhatsApp — no separate triage form.
  2. 2
    Decision. the agent splits the message into separate concerns, scores each against the program's clinical-acuity policy and the patient's memory (medication, titration stage, what they have already reported), and orders them — self-manage, watch, or escalate — with conservative defaults that round any ambiguity upward.
  3. 3
    Action. low-acuity concerns get calm, approved self-care guidance and what-to-watch-for in the same reply; the highest-acuity concern is routed to a clinician first as a ranked, sourced summary with the lower-acuity items listed beneath it so nothing is lost.
  4. 4
    Follow-through. the ranked triage decision is logged to the patient record; the clinician opens the queue already sorted by urgency; a "this got worse" reply re-ranks and re-escalates immediately, and a judge reviews every unprompted message before it sends.
Example
PatientA few things — feeling a bit queasy since my shot, the injection spot is sore and a little red, and I keep getting this tight feeling in my chest when I walk upstairs.
AgentThanks for laying all three out. The queasiness and a sore, slightly red injection site are common and usually settle — I've added what to watch for below. The chest tightness on exertion is the one I won't sit on: I've flagged it to your care team as urgent with a summary of all three, and they'll reach you shortly. If it worsens or you feel breathless before then, treat it as an emergency.

Configuration

How the agent is wired for this use case.

Triggeran inbound patient message describing one or more symptoms (symptom_report) on chat, in-app, or WhatsApp, optionally drawing on recent logged reports.
Tools & actions
  • Messaging channel · receive the report, return approved self-care guidance for the low-acuity concerns, and tell the patient what was escalated
  • Patient memory / history · read medication, titration stage, and prior reports to weigh each concern in context
  • Knowledge base · retrieve the program's approved self-care guidance and the watch-for / escalate-if criteria per symptom class
  • Clinician escalation queue · file a ranked, sourced summary with the highest-acuity concern surfaced first and the rest listed beneath
  • Care-record system · log the parsed concerns, the acuity ranking, and the routing decision for the record and audit
Autonomyparsing, acuity ranking, and low-acuity self-care guidance run unattended under a clinical-triage policy with conservative defaults — when unsure, escalate. The agent gives no diagnosis, dosing, or schedule advice; it ranks urgency and routes only, and every message is judge-gated. Any high- or uncertain-acuity concern is a mandatory clinician hand-off.
Channelschat · in-app · whatsapp
Escalationany concern scored high or uncertain acuity, or a worsening follow-up, routes to a clinician first with the full ranked summary attached; automated replies stand down on that concern once a clinician owns it.

What you need

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

Signals

inbound multi-symptom messages, recent logged symptom reports, patient follow-up replies.

Data

patient history, medication and titration stage, prior symptom reports, persistent memory, consent state, the program's approved guidance and per-symptom acuity criteria.

Guardrails

the agent ranks acuity and routes only — no diagnosis, dosing, or schedule advice ever; conservative defaults with mandatory clinician hand-off on any high or ambiguous concern; judge gating on every message; a clear scope boundary stated to the patient; full audit trail of every triage and ranking decision.

Metrics it moves

  • safe-escalation-rateup, as the one concern that matters inside a multi-symptom message is reliably surfaced first instead of buried behind routine ones.
  • time-to-resolutiondown, because the clinical queue arrives pre-ordered by acuity, so urgent cases are seen before harmless ones.
  • contact-ratedown, since low-acuity concerns are resolved with approved guidance in the same reply rather than each consuming a clinical slot.

Related use cases

Prevent / Care

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

the single-symptom detect-and-route flow this layer sits above

Prevent / Care

Clinical red-flag detection and safe escalation

the binary red-flag rail this generalizes into ranked acuity

Prevent / Care

Async message-urgency triage for the clinician inbox

the clinician-side counterpart that orders the existing inbox

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