All use cases

Onboard / Activate

Contraindication and boxed-warning intake screen

The patient who should never be offered a GLP-1 medication is best caught at intake — not after a prescription is already in motion.

Next best thingSystem triggerBehavioral triggerChatIn-appGLP-1 & PeptidesHealth & Telehealth

What it does

GLP-1 medications carry hard contraindications: a personal or family history of medullary thyroid cancer or MEN2, a history of pancreatitis, and pregnancy or trying to conceive. In a fast conversational intake those answers are easy to skim past, and a contraindicated patient who slips through is a serious safety and liability event. The agent runs a dedicated contraindication layer inside intake: it asks the boxed-warning questions in plain language, recognizes a positive answer, and surfaces it for the prescriber before any medication is offered. It screens and flags; it never clears a patient or rules one out on its own — that decision stays with the clinician.

How it works

  1. 1
    Trigger. a patient reaches the screening step of a GLP-1 intake (in chat or in-app), or gives an answer mid-conversation that touches a contraindication, so the screen runs in-session as the conversation unfolds.
  2. 2
    Decision. the agent works through the contraindication set from the medication's labeling — thyroid-cancer or MEN2 history (personal or family), prior pancreatitis, pregnancy or conception intent — interpreting plain-language answers, asking a clarifying follow-up where an answer is ambiguous, and classifying each as clear, positive, or needs-clinician-review.
  3. 3
    Action. a positive or ambiguous answer is flagged to the prescriber with the patient's exact words attached, and the patient is told plainly that a clinician needs to review this before anything goes further — not given a yes or no on eligibility.
  4. 4
    Follow-through. the screen and its flags write back to the patient record as a structured, auditable result; the clinician makes the call; the agent gives no clinical opinion on whether the medication is safe for this patient.
Example
AgentA couple of safety questions before we go further. Has anyone in your family ever had medullary thyroid cancer, or a condition called MEN2?
CustomerMy mum had thyroid cancer, I'm not sure which kind.
AgentThank you for telling me — that is exactly the kind of detail a clinician needs to look at before anything is prescribed. I've flagged it for your care team to review with you, and they'll be in touch. I'm not able to give a yes or no on the medication myself.

Configuration

How the agent is wired for this use case.

Triggerthe patient reaches the screening step of a GLP-1 intake, or gives an in-conversation answer that touches a contraindication — the screen runs in-session via an intake_screening_step event.
Tools & actions
  • Messaging channel · ask the boxed-warning questions in plain language and run a clarifying follow-up on any ambiguous answer.
  • Knowledge base · sequence the contraindication set from the medication's labeling (thyroid-cancer / MEN2, pancreatitis, pregnancy) and the screening logic.
  • Clinical/EHR system · flag a positive or ambiguous answer to the prescriber with the patient's exact words, and write the screen result back as a structured, auditable record.
  • Clinical escalation path · hold the candidate for clinician review rather than returning an eligibility decision.
Autonomyasking, interpreting, flagging, and write-back run unattended within the labeled contraindication logic; under a detect-and-escalate clinical policy the agent screens and flags only — it never clears a patient, never rules one out, and gives no opinion on whether the medication is safe; every screen result is logged for audit.
Channelschat · in-app
Escalationany positive or ambiguous contraindication answer holds the candidate for prescriber review before a medication is offered.

What you need

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

Signals

intake_started, the intake_screening_step event, and per-question answer events from the intake flow.

Data

the labeled contraindication set and screening logic, the patient's intake answers, consent state, the patient record to write flags back to.

Guardrails

detect-and-escalate clinical policy — the agent screens and flags, never clears or diagnoses; any positive or ambiguous answer routes to a clinician; the contraindication logic is clinician-owned and version-controlled; every screen result logged for audit.

Metrics it moves

  • intake-completionup, because boxed-warning questions answered conversationally with a clarifying follow-up are skipped far less often than the same items buried in a static form.
  • contact-ratedown on avoidable safety escalations later, by catching contraindicated candidates at the screening step instead of after a prescription is in motion.

Related use cases

Onboard / Activate

Conversational intake and eligibility check

the broader guided intake this safety layer sits inside

Onboard / Activate

Baseline-labs chase before first injection

the next pre-prescribing gate once screening is clear

Prevent / Care

Pregnancy and conception washout safety stop

the same contraindication enforced later in the journey

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