Pearl, the 24/7 AI receptionist

The challenge

Pearlline is a fictional four-chair dental chain in Bengaluru. Like most clinics, its front desk is busiest exactly when patients call: during treatment hours. Calls go unanswered, after-hours enquiries leak to competitors, and reception staff spend their day repeating timings, prices and directions.

Our approach

We mapped the 12 call types that make up most inbound volume — new bookings, reschedules, pricing questions, insurance, emergencies and directions — and designed a voice persona, Pearl, that handles the routine ones end-to-end and hands off the rest with context.

• Conversation design: short turns, interruption handling, Hindi–English code-switching, and a confirmation step before anything is booked.
• Guardrails: Pearl never gives clinical advice. Pain, swelling or bleeding triggers an immediate transfer to the on-call number.
• Calendar truth: availability is read live from the practice calendar, so Pearl can't double-book a chair.

The solution

Calls land on a Twilio number and stream to a low-latency voice pipeline. Claude handles dialogue and tool use (check slots, book, reschedule, send a reminder), an n8n workflow writes the visit to the calendar and patient sheet, and a WhatsApp confirmation goes out within seconds. Every call is summarised for the front desk.

Projected impact

These are design targets for a concept build, not client results: answer 100% of calls including nights and Sundays, handle roughly two-thirds of bookings without staff, and free 2–3 hours of front-desk time per day.