AI Voice Agent for Patient Discharge Calls
An AI voice agent that calls discharged patients, runs a check-in questionnaire in natural conversation, and escalates red flags to a nurse.
The workflow
The complete flow, exactly as it runs in the Workflow Builder. Scroll to explore, or use the zoom controls.
Requirements
Set up these connections in the Workflow Builder before importing or running this workflow.
Provides the SMART on FHIR embed context used to identify the discharged patient and write questionnaire results back to the chart.
Powers the outbound voice call. The workflow dials the patient over Twilio and streams the conversation to the real-time AI for live transcription and response.
About this workflow
After a patient leaves the hospital, someone should check in on them — but there's never enough staff to call everyone. This workflow is an AI voice agent named “Beth” who actually phones discharged patients, walks them through a check-in questionnaire in natural, warm conversation, captures their answers as structured data, and flags anyone whose responses say they need a nurse.
The problem
Post-discharge follow-up calls are one of the highest-value things a care team can do — they catch complications early and measurably reduce readmissions. But they're slow, repetitive, and the first thing to fall off the list when staff are stretched. The result is that many patients, often the ones who most need a check-in, never get the call.
How it works
An API kicks off an outbound call (over Twilio) and connects it to a real-time conversational AI. Beth opens by verifying she's talking to the right person, then asks the questionnaire one question at a time — rephrasing clinical wording into plain language and never sounding scripted.
Behind the voice, a second AI agent parses each spoken reply into a clean, structured answer: a 1–10 score, a yes/no, a multiple-choice option. A rule engine checks each answer against red-flag thresholds, and the instant something crosses the line — a high pain score, a fever, a worrying symptom — it escalates that patient to the care team.
The call history and questionnaire results are kept per call, ready to be reviewed or written back.
Why it matters
Every discharged patient can get a timely, consistent check-in — in a real conversation rather than a robotic survey — and the answers come back structured and ready for the chart instead of trapped in a voicemail. Crucially, staff attention is routed by the red-flag engine to the patients who actually need a human, instead of being spent dialing through a list.
Who it's for
Care-management, oncology, and post-acute teams who want to scale follow-up outreach without adding headcount, and product teams exploring voice AI for patient engagement and remote symptom monitoring.
How to build this workflow
Recreate it in the Workflow Builder by wiring these steps in order — or download it above and import it directly.
Step 1: Receive the Request
A trigger starts the agent — a chat question, an API call, or an inbound/outbound phone call.
Step 2: Authenticate & Set Context
Sign in to the EHR and load the context the agent needs — the patient, the conversation so far, and the tools it can use.
Step 3: The Agent Reasons
An LLM-powered agent interprets the request and decides what to do next.
Step 4: Use Tools & Take Action
The agent calls tools — searching FHIR data, asking the next question, placing the call — and loops until it has what it needs.
Step 5: Guardrails & Safety
Output passes through guardrails — PHI redaction, allowed-resource checks, red-flag escalation — before anything leaves.
Step 6: Respond
The agent returns its answer — text, a spoken reply, or a follow-up action like alerting a nurse.