Setup checklist for a medical AI receptionist

A complete checklist for deploying the 365agents Medical Office agent — EHR integration, HIPAA BAA, clinical protocols, nurse-line triage configuration, and provider scheduling rules.

Written By Rick Garcia

Last updated 15 days ago

Deploying an AI receptionist in a medical office takes more preparation than most industries because of HIPAA requirements, EHR integration, and clinical protocol configuration. Even so, most primary care practices are fully live within 2–3 weeks of kickoff. This checklist walks through exactly what to prepare and what to expect.

Before the kickoff call

  • EHR / practice management system — which one (Epic, Cerner/Oracle, Athena, eClinicalWorks, NextGen, ModMed, Kareo, Practice Fusion) and admin credentials for scheduling API access
  • Provider roster — all providers with their specialties, new-patient availability, and individual schedule rules
  • Practice hours and holiday schedule — by location if multi-location
  • Service list — what visit types you offer (new patient, established, annual, sick, specific procedures)
  • Accepted insurances — in-network carrier list and plan types; self-pay rates where relevant
  • Nurse-line triage protocols — your current scripts for fever, pain, respiratory symptoms, medication questions, etc.
  • After-hours coverage — who's on call, how they're reached, escalation rules
  • Emergency escalation rules — what gets routed to 911 guidance vs. nurse vs. provider
  • Prescription refill workflow — who reviews and approves refills, how long it normally takes, pharmacy preferences
  • Top 20–30 FAQs — what does your front desk answer most often

Decisions you'll need to make

  • Voice — we default to Aria (calm, professional) for medical
  • Disclosure approach — we recommend proactive disclosure; confirm wording
  • After-hours handling — AI all hours, or AI only after hours?
  • Clinical scope — exactly what clinical questions (if any) you want the AI to answer from the knowledge base vs. always route to a clinician
  • New-patient intake — how much information collected before booking
  • Lab result handling — confirm AI only tells patients results are "available" and does not summarize
  • Provider preferences — some providers prefer the AI collect more context; others prefer minimal pre-call info

HIPAA setup

  • Countersign our HIPAA Business Associate Agreement
  • Identify authorized personnel who can access call recordings and transcripts
  • Configure PHI retention period (30 / 90 / 365 days or longer per your policy)
  • Review our HIPAA-specific configuration checklist with the onboarding team
  • Update your Notice of Privacy Practices to include the AI as a Business Associate if required by your state

EHR integration

  • Epic — integration via MyChart scheduling APIs and/or USCDI-compliant FHIR
  • Cerner / Oracle — via their scheduling API
  • Athena, eCW, NextGen, ModMed, Kareo — via their respective scheduling and patient-lookup APIs
  • Smaller / proprietary systems — we work with most; ask during kickoff
  • Read-only starter — some practices deploy the AI with calendar-read-only first, manually syncing bookings, until EHR write-back is validated

Call forwarding and routing

  • Your current phone provider (or cloud PBX) — we have setup guides in the Call forwarding collection
  • Decide on forwarding rules (always, after-hours, overflow, no-answer)
  • Confirm E911 configuration for VoIP lines
  • Set up the nurse-line transfer number and on-call rotation routing

Clinical protocol configuration

  • Triage protocols — your existing scripts become the AI's decision tree
  • Urgent-symptom list — what triggers immediate clinical escalation
  • Emergency-symptom list — what triggers 911 guidance
  • Medication-specific routing — controlled substances, medication adjustments requiring provider approval
  • Pediatric-specific rules if you see kids

Go-live sequence

  • Week 1: kickoff, requirements, EHR integration, HIPAA BAA, draft agent instructions
  • Week 2: knowledge base loaded, clinical protocols configured, internal test calls
  • Week 3: soft launch (after-hours or overflow), clinical team feedback, tuning
  • Week 3–4: full rollout, daily transcript review for the first two weeks

What to measure after go-live

  • Call containment rate by call type
  • Nurse-line escalation accuracy (are the right calls reaching the nurse?)
  • Emergency escalation time (how fast do urgent calls reach a clinician?)
  • New-patient capture rate
  • Refill processing time
  • Patient satisfaction sampling

Related articles

Start the process

Book a demo and our healthcare onboarding team will walk through this checklist with you. Most primary care practices leave the initial call with a clear 2–3 week go-live plan.