Automation blueprint · Clinics & doctors · 8 min read
The complete automation blueprint for a 2-doctor Bengaluru clinic.
Every repetitive workflow at a typical OPD front desk, inventoried hour by hour — and the exact system that removes them. Take it to any developer; the math is the same.
This is a modeled clinic, not a client — yet. Every number traces to our published assumptions at conservative tier-2 salary rates, rounded down. When our first clinic goes live, this page becomes their (anonymized) case study, measured, not modeled.
The clinic
- Doctors
- 2 (OPD, six days a week)
- Front office
- 2 staff
- Patient inquiries
- ~30/day, phone + WhatsApp
- Appointments
- ~25/day
- No-show rate
- ~15% of bookings
- Reports going out
- ~8/day
A day at the front desk
8:45 AM
Phones start before the shutters are fully up.
“Aaj Dr. Mehta hain?” — the same three questions, thirty times a day.
9:30 AM
The reminder-call list from yesterday's register.
Twenty-five confirmations, three minutes each, every single morning.
12:40 PM
“Is my report ready?” — call number four on that report.
Eight reports a day, each generating two or three status calls.
3:15 PM
Two no-shows from the morning. Nobody has time to chase them.
Empty chair time that a waitlist could have filled.
7:50 PM
Closing up. The six-month recalls didn't happen today either.
The most profitable message a clinic can send is the one nobody sends.
The repetitive work, inventoried
Conservative estimates from our published assumptions, rounded down — and built so no hour is ever counted twice across automations.
| Task today | Automation | Hrs/mo |
|---|---|---|
| Answering routine questions (timings, fees, directions) | AI WhatsApp receptionist ● high | 31 |
| Booking & rescheduling coordination | In-chat appointment booking ● high | 20 |
| Morning reminder calls | Reminder ladder (T-24h / T-2h) ● high | 29 |
| “Is my report ready?” calls | WhatsApp report delivery ● high | 10 |
| Chasing no-shows | Same-day rebooking recovery ● high | 5 |
| Recall outreach that never happens | Check-up recall campaigns ● high | 4 |
| Total repetitive work found | 99 | |
The system
Five pieces, all standard, all replaceable. The clinic owns everything.
- WhatsApp Business API
The front door. Patients never install anything new.
- Claude
Understands Hinglish, answers from the clinic's own approved facts, and knows exactly what it may not answer.
- n8n
The nervous system: routes messages, fires reminders, writes records, escalates to humans.
- Google Calendar
The clinic's existing calendar stays the source of truth.
- Supabase
The clinic's own database — patient data lives in their instance, not ours. DPDP-conscious by architecture.
Build order & what it costs
The structure is fixed; we publish exact numbers the moment our first client conversations confirm they're fair — the structure won't change, and your audit roadmap conversation includes the current figures.
Phase 1 · The AI employee
Weeks 1–2
WhatsApp receptionist with in-chat booking (A1 + B1 — one integrated deployment on the same bot)
One-time setup + monthly retainer
Goes live on the clinic's own number. Every behavior approved by the doctors before patients see it.
Phase 2 · Quick wins
Week 3
Reminder ladder (B2) + report delivery (CL3)
2 × One-time
Set up once, run themselves. No retainer needed for these.
Phase 3 · Optional
Week 4, if wanted
No-show recovery (B3) + recall campaigns (CL4)
2 × One-time
Adds ~9 hrs/month and recurring-visit revenue. Fine to skip at first.
The math, conservatively
~20 hours a week back.
The recommended build (Phases 1–2) removes ~90 staff-hours/month — worth ~₹6,300/month in salary time even at conservative tier-2 rates. That’s the bar any retainer we quote has to clear on reclaimed hours alone — before counting a single recovered no-show, filled slot, or midnight booking. We’d rather the upside surprise you than the invoice.
What stays human
- Every medical answer. The AI books, reminds, and informs — it never advises.
- Report release: the doctor approves every report note before it goes out.
- Fees and exceptions: pricing questions beyond the published list go to a human.
- The schedule itself: doctors set the structure; the AI fills it.
Steal this blueprint. Genuinely.
Take it to any developer — the architecture and the math hold. Or run the audit and get the version personalized to your clinic’s actual numbers in ten minutes.