AI front desk handling 4,000 patient queries a month
A multi-speciality network deployed an AI assistant for appointments, reports and insurance questions — with clinicians in the loop.
Client Feedback
“It answers the boring questions perfectly and knows exactly when to fetch a human. That balance was everything.”
Dr. Kavita Menon
Medical Superintendent, Aarogya Health Network
Measurable business results
78%
Queries fully automated
-92%
Average wait time
4,000
Monthly conversations
0
Clinical advice incidents
The Problem
Reception handled 4,000 repetitive queries monthly. Average hold time was 4m 20s and 23% of callers hung up.
Before our intervention:
- 4m 20s average hold time
- 23% call abandonment
- Reception overloaded 10am–1pm
The Outcome
Web and WhatsApp channels launched together, with a human handoff SLA of 90 seconds.
After launch:
- Instant first response
- 6% escalation to humans
- Reception focused on in-clinic patients
How we designed and engineered it
Discovery & Insights
We clustered 12,000 call transcripts into 41 intents. Just 9 intents covered 78% of all contacts.
UX & System Design
A calm, text-first assistant with explicit escalation to a human, and never a medical opinion — hard guardrails written with the clinical team.
Engineering & Quality
Retrieval over an approved knowledge base, appointment API integration, full transcript logging and a clinician review queue.
Services Delivered
Technology Stack
Related case studies
Booking-first website for a 9-clinic dental group
A multi-location dental group was losing patients to a slow, unbookable website. We rebuilt it around one action: book an appointment.
View projectShop-floor ERP that replaced 27 spreadsheets
A precision components manufacturer ran production on spreadsheets and WhatsApp. We built one system for job cards, inventory and dispatch.
View project