How a Pune Clinic Booked 40 More Appointments Per Week with WhatsApp AI

The Problem

Sunrise Multi-Specialty Clinic in Baner, Pune was drowning in WhatsApp messages. Three full-time receptionists were each handling 100+ messages per day — enquiries about appointments, doctor availability, test reports, directions, and everything in between.

The math was brutal: ₹30,000/month in salary, plus ₹10,000 in overhead, for conversations that were 70% repetitive.

The Solution

We deployed the SpycAI WhatsApp Sales Agent in January 2026. The setup took 11 days from first meeting to go-live. We built a knowledge base from:

• ✓3 years of WhatsApp conversation logs (anonymised)
• ✓Doctor schedules and availability rules
• ✓Service catalogue with pricing
• ✓Location and navigation FAQs

The Results (30 Days)

The 71% resolution rate meant the clinic's three receptionists could focus on in-person patients instead of their phones. Staff satisfaction actually improved.

Lessons Learned

Hindi/English code-switching matters. Standard English LLMs failed at queries like "doctor ka time kab hai" mixed with "can I book 3pm slot?" We implemented Sarvam AI for Marathi queries and fine-tuned our Hinglish prompts extensively.

The handoff protocol is critical. When the AI is unsure, it must hand off gracefully — not silently fail. We built a confidence threshold of 0.7: below that, the AI says "Let me connect you to our team" and pings Slack.

Templates need WhatsApp approval. Campaigns require pre-approved templates from Meta. We now maintain a library of 40 approved templates for the clinic.

Conclusion

For a clinic with 300+ daily WhatsApp messages, the ROI was clear in week one. The agent paid for itself in saved receptionist hours within 11 days of deployment.