Singapore · Clinic operations & health screening

I ran a clinic's appointment book for four years. Then I built the system that handles it.

3,215 patient conversations since March — 94% answered without anyone stepping in. It's live on healthscreening.sg right now. Go and try it.

Start with a map See what's running

3,215

Patient conversations handled9 March – 18 July 2026

94%

Answered without a human195 needed the clinic team

78%

Of screening-package revenue books through the systemup from 42% the month before

65%

Of conversations are sent a booking link2,103 of them

Counted from the assistant's own event log, 9 March – 18 July 2026. “Answered without a human” means no one from the clinic joined the conversation. Revenue figures are reconciled against issued invoices, pre-GST. One client: a Singapore health-screening group running four locations.

Why me

Forty to seventy WhatsApp conversations a day. By hand.

That was my job for four years — coordinating appointments for a Singapore health-screening group across four clinics. Quoting from a catalogue of over a thousand services. Checking the practice-management system for slots. Typing bookings in one at a time.

So when I built the system that now does most of it, I wasn't guessing at the workflow. I'd been the workflow. I know which questions patients actually ask, which ones need a person, and which parts of a clinic's day look automatable but aren't.

That's the difference between a product built for clinics in general and a system cut to fit yours. A dovetail joint holds because each piece is cut to the shape of the other — no nails, no forcing. Same idea.

What's running right now

Two halves of one system.

The assistant patients talk to

On the clinic's own WhatsApp number and website. It answers from a live catalogue of 1,062 services and prices, checks real availability, and takes the patient through to a booking — then hands over to the clinic team when it should.

94% of conversations resolved without escalation, improving every month since launch.

The integration layer underneath

The part nobody demos. Middleware that talks to Plato, the clinic's practice-management system — reading availability, creating and rescheduling appointments, reconciling bookings against issued invoices, and syncing home-screening visits out to an external phlebotomist's calendar.

Where a generic chatbot stops, this is where the work actually is.

What you don't have to change

Nothing gets ripped out.

Keep Plato. The system works with it, not instead of it.
Keep your clinic's existing WhatsApp number.
No new logins for your front desk to learn.
No data migration, and no new system of record.
Your team still handles anything the assistant shouldn't.
Every action is logged, and the logs are yours.

Where to start

Start with a map, not a build.

Most clinics don't need everything automated. They need to know which two or three things are worth it — and which ones will cost more than they save. That's what this is.

The Clinic Operations Map

Ninety minutes with whoever actually runs your day — walking your real workflows, not a questionnaire.
A written map within five working days — your workflows ranked by what automating them would actually be worth.
An explicit list of what not to automate — and why. In a clinic, the boundaries matter more than the opportunities.
A straight answer on AI accounts and PDPA — whether your team needs a commercial plan, and what happens to patient data if staff are using personal ones. Often the answer is that you don't need to buy anything.

S$1,500 for a single location, S$2,500 for a multi-site group — credited against a build if we go ahead.
Free for my first three clinics, while I'm building case studies. You keep the map either way, including if it concludes you shouldn't build anything.

Ask about a map

If you might want government grant support for a build later, talk to me before you sign anything or pay a supplier. Starting work too early can disqualify a project, and a Letter of Offer takes eight to twelve weeks.

Get in touch

I'll tell you honestly whether it's worth building.

If something in your clinic's day is eating hours, or you're being pitched AI and want a second opinion from someone who's actually shipped it — tell me about it.