Client
MediAI Labs
Industry
Healthcare
Tech Stack
TwilioOpenAI RealtimeNode.jsEpic FHIRReact NativePostgreSQL
The Challenge
What We Were Up Against
MediAI's four clinics missed 40% of inbound calls outside business hours. Each missed call represented an average $280 lifetime patient value. Staff spent 3 hours daily on scheduling — time better spent on patient care.
Our Solution
How We Solved It
We deployed a custom voice AI agent using Twilio and OpenAI's Realtime API that handles calls end-to-end: answers in <1s, understands natural speech, checks live appointment availability, books, confirms via SMS, and escalates clinical questions to staff.
01
Call Flow Analysis
Recorded and transcribed 2,000 inbound calls to build an intent taxonomy of every query type.
02
Voice Agent Build
Twilio Voice + OpenAI Realtime API with a custom conversation state machine for healthcare scheduling logic.
03
EHR Integration
Real-time availability from Epic via HL7 FHIR API — agent sees actual appointment slots, not a cached copy.
04
HIPAA Compliance
BAA with all vendors, data encryption at rest and in transit, no PHI stored in AI context.
05
Staff Escalation
Clinical or distressed-caller detection triggers instant warm transfer to on-call staff with transcript context.
The Impact
Measurable Results
60% growth in new patient bookings within 90 days
Missed calls dropped from 40% to 8%
24/7 availability with <1 second answer time
3.1× booking conversion vs. voicemail
Staff scheduling time reduced by 3 hours/day
"Patients tell us the AI is 'so easy to talk to'. We've grown 60% without adding a single receptionist — and the staff love not answering scheduling calls all day."
Dr. Aisha Patel
Founder, MediAI Labs
Key Outcomes
New patient growth60%
Availability24/7
Missed calls−52%
Booking conversion3.1×
