All use cases

Engage / Lifecycle

Peptide reconstitution and dosing-math assist

A patient is standing at the counter with a powder vial, a bottle of bacteriostatic water, and a chart they half-understand — the agent walks their exact prep steps so the concentration math is right before anything goes in the syringe.

Next best thingBehavioral triggerTime-based triggerChatIn-appGLP-1 & PeptidesHealth & TelehealthWellness (B2C apps)

What it does

A lyophilized peptide does not arrive ready to inject — the patient has to reconstitute the powder with bacteriostatic water, then read the supplied chart to convert their prescribed dose into syringe units. This is where it breaks: mix in the wrong volume of water or misread the units line and the same powder becomes a dose that is off by a multiple, and the patient does it alone, at home, the first time. Nobody is in the room at that moment, so the error either lands in the body or turns into an anxious support ticket. The agent walks the patient through the exact reconstitution and units math for their vial and their chart, catches the classic concentration mistake before the draw, and routes anything that does not match the protocol to the clinical team.

How it works

  1. 1
    Trigger. the patient opens a "how do I mix this / how many units" conversation in chat or in-app, or a time signal fires on a known first-prep or vial-change moment, before the dose is drawn.
  2. 2
    Decision. the agent pulls this patient's vial strength, prescribed dose, and the reconstitution chart on file from memory and the knowledge base, then works their specific numbers — how much bacteriostatic water the chart calls for and which units mark on the syringe matches their dose. It checks the answer against the protocol rather than improvising one.
  3. 3
    Action. it walks the steps in order on the patient's channel — draw this much water into the vial, swirl do not shake, let it dissolve, then your dose sits at this units mark — and reads the math back so the patient can confirm before injecting.
  4. 4
    Follow-through. the confirmed prep writes back to memory so the next vial is faster; if the patient's numbers do not match the supplied chart, the dose is unusual, or they describe a different vial than the record, the agent stops the math and hands to the clinical team with the details rather than guessing. Sends are judge-gated and measured against a holdout.
Example
PatientI've got the powder and the water. I don't get how many units my dose is.
AgentLet's set it up for your vial. Read me the water amount your chart lists for this strength and I'll confirm it back — then draw exactly that into the vial slowly down the side, swirl gently until it's clear, don't shake. Once it's mixed, your prescribed dose sits at the marked line on the chart for your syringe. Want me to walk that line back so you can match it before you draw?
PatientMine says a different water amount on the box than what you said.
AgentThen let's not guess at the concentration. I'm passing this to your clinical team with your vial and chart details so they confirm the right mix for you before you inject — I'll let you know the moment they reply.

Configuration

How the agent is wired for this use case.

Triggera reconstitution or units question raised by the patient in chat or in-app, or a first_prep_due / vial_change time signal from the app backend, fired before the dose is drawn.
Tools & actions
  • Knowledge base · supplies the manufacturer reconstitution chart, dilution table, and prep steps for this vial strength.
  • Patient memory store · reads the vial strength, prescribed dose, and chart on file; writes back the confirmed prep so the next vial is faster.
  • Clinical / EHR system · confirms the active prescription and dose of record so the math is run against the real order.
  • Messaging channel · walks the ordered steps and reads the units math back for confirmation on chat or in-app.
  • Clinician escalation queue · receives any chart mismatch, unusual dose, or wrong-vial report with the captured details.
Autonomythe agent reads back the patient's own protocol chart and the supplied prep steps under judge gating; it never computes a new dose, changes the prescribed dose, or improvises a concentration the chart does not list — any deviation between the patient's numbers and the chart of record is detected and escalated to a clinician, not resolved by the agent.
Channelschat · in-app
Escalationa chart-versus-record mismatch, an out-of-protocol dose, a described vial that differs from the prescription, or any safety doubt hands off to the clinical team with the prep details.

What you need

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

Signals

the patient-initiated reconstitution / units question, a first-prep or vial-change event from the app backend.

Data

vial strength and prescribed dose on file, the manufacturer reconstitution chart and dilution table, patient memory of prior preps, channel consent.

Guardrails

the agent only reads back the patient's supplied chart and protocol; no dose-setting, no new concentration the chart does not list; any mismatch or unusual dose escalates to a clinician (detect, then escalate); judge gating on every send; holdout measurement.

Metrics it moves

  • adherenceup, because a patient who can prep the vial correctly the first time actually takes the dose instead of stalling on the powder.
  • contact-ratedown, the "how do I mix this / how many units" question is answered in the moment rather than becoming a support ticket or a missed dose.
  • csatup, the scariest at-home step is guided instead of guessed, and a genuine mismatch reaches a clinician fast.

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