HealthFirst Patient Bot
HealthFirst Clinic, Chennai
Overview
HealthFirst is a multi-specialty clinic in Chennai with 12 doctors and 200+ patient interactions daily. Their administrative staff was overwhelmed handling repeat inquiries, scheduling, and sending reminders — all manually via WhatsApp.
The Challenge
The clinic's team of 3 reception staff spent over 3 hours every day answering the same questions: 'What are your timings?', 'Is Dr. X available?', 'How do I reschedule?'. On top of that, appointment no-shows were at 35% because reminders were sent inconsistently when staff remembered.
Our Solution
We built a GPT-4 powered WhatsApp bot connected to the clinic's appointment system via Calendly API. The bot handles: initial patient greeting and service information, available doctor schedules and specialty routing, appointment booking with real-time slot checking, automated reminders 24h and 1h before appointments, rescheduling and cancellation flows, and FAQ handling for 40+ common questions. For complex queries, it seamlessly transfers to a human staff member with full conversation context.
The Outcome
Within 30 days of launch: appointment no-shows dropped from 35% to 14%, staff freed up 3+ hours daily which was redirected to patient care, 78% of all inquiries handled without human involvement, and patient satisfaction scores improved from 3.8 to 4.6 out of 5.
How we delivered it
Discovery (Days 1–3)
Audited 3 months of WhatsApp messages to identify the top 40 questions. Mapped appointment workflow and CRM data structure.
Bot Training (Days 4–8)
Trained GPT-4 on clinic's service catalog, doctor profiles, FAQs, and policies. Built intent detection for appointment vs inquiry vs complaint flows.
API Integration (Days 9–14)
Connected WhatsApp Business API via WATI, integrated Calendly for real-time slot management, and built PostgreSQL log for conversation history.
Testing & QA (Days 15–18)
Ran 200+ test conversations including edge cases. Staff tested handoff triggers and override controls.
Launch & Monitor (Days 19–21)
Soft launch with 20% of incoming messages, monitored fallback rate, fine-tuned responses, then full rollout.
Tech Stack
Want similar results?
Let's talk about what we can build for your business.