All use cases

Prevent / Care

Gray-market peptide safety intercept

A member asks how to add a "research-only" vial they found online — the agent declines to coach it and puts a clinician between them and the bottle.

Get aheadNext best thingBehavioral triggerRisk triggerChatIn-appGLP-1 & PeptidesHealth & TelehealthWellness (B2C apps)

What it does

Peptide-therapy patients sit one search away from a gray market of "research use only" vials sold with no guarantee of purity or dosing. When a member raises sourcing an unlisted compound elsewhere, stacking an over-protocol dose, or buying outside the program, a generic assistant either answers the question helpfully or ignores the risk — both of which leave the patient alone with an unverified bottle and the program carrying the liability. The agent treats unsafe-sourcing intent as its own risk signal, separate from any symptom: it recognizes the intent, declines to instruct or validate it, and routes the member to the prescribing clinician with the context attached. It never coaches a non-prescribed compound and never supplies a dose for one.

How it works

  1. 1
    Trigger. a member message that signals sourcing or dosing outside the program — a "research-only" vendor, an unlisted or non-prescribed compound, or an over-protocol dose — surfaced behaviorally from the conversation, not from a form.
  2. 2
    Decision. the agent classifies the intent against a sourcing-risk policy rather than a symptom rail; an off-program compound or over-protocol ask is a hard non-coach, distinct from a routine technique or side-effect question it would normally handle.
  3. 3
    Action. it declines to instruct or affirm the choice, gives only bounded "your clinician needs to weigh in before anything like this" language, and opens a routed handoff to the prescriber with the member's request captured.
  4. 4
    Follow-through. the intent and the handoff write back to the member's memory so the pattern is visible to the care team; the clinician's response closes the loop, and the agent never resumes coaching the compound on its own.
Example
MemberI found a peptide vial online labelled research-only that's way cheaper — can you tell me how much to draw with what I'm already on?
AgentI can't help dose or combine something sourced outside your program — those vials aren't quality-checked, and your clinician has to look at it against what you're already taking. I've flagged this to your prescriber with your question so they can advise you directly.
MemberOkay, makes sense. I'd rather be safe.

Configuration

How the agent is wired for this use case.

Triggera member message signalling off-program sourcing or dosing (a "research-only" vendor, an unlisted or non-prescribed compound, an over-protocol dose), detected behaviorally in the conversation
Tools & actions
  • Messaging channel · recognize the sourcing-risk intent and reply with bounded, non-coaching language
  • Knowledge base · match the request against the program's approved-compound and protocol list to confirm it is off-program
  • Clinical/EHR system · open a routed handoff to the prescribing clinician with the member's request attached
  • Memory store · log the intent and the handoff against the member's history for the care team
Autonomydetect and route only — the agent runs unattended to recognize the intent and stand down, under a sourcing-risk policy class; it never instructs, validates, or doses a non-prescribed compound, and every message is judge-gated. Detect, then escalate to a clinician; no dose advice.
Channelschat · in-app
Escalationany off-program sourcing or over-protocol intent hands to the prescribing clinician; automated coaching stands down on that compound until the clinician owns it

What you need

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

Signals

in-conversation mentions of off-program sourcing, unlisted compounds, or over-protocol dosing

Data

the program's approved-compound and protocol list, the member's prescribed regimen, persistent member memory, consent state

Guardrails

sourcing-risk policy with conservative defaults; the agent never instructs, doses, or affirms a non-prescribed compound; detect-then-escalate only, no dose advice; judge gating on every message; full audit trail of the intent and the handoff

Metrics it moves

  • safe-escalation-rateup, as unsafe-sourcing intent reliably reaches a clinician instead of being answered or ignored
  • contact-ratedown on the volume the clinician would otherwise spend untangling an off-program request after the fact
  • csatup, since the member feels protected rather than either enabled or stonewalled

Related use cases

Prevent / Care

Clinical red-flag detection and safe escalation

the symptom-side rail this sourcing-intent guard sits beside

Prevent / Care

Hematocrit and blood-pressure safety-flag escalation (TRT)

a physiologic safety halt in the same peptide audience

Prevent / Care

Off-label request escalation guard

the prescribing-boundary guard for off-label asks

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