Clínica BC — An integrated AI + management system

The problem

The clinic ran on the doctor's personal WhatsApp. Every new-patient message landed in the same inbox as her private life — so it became chaos: messages got buried, some were never seen, and replies often came hours (sometimes a day) late.

For an aesthetics clinic, that slow first response is lost revenue. A prospective patient who doesn't hear back quickly books somewhere else. The doctor was also doing triage and scheduling by hand, on top of actually seeing patients — so the bottleneck was her, and it capped how many new clients the clinic could take on.

Two problems, really:

— Front of house was slow, manual, and leaking leads.

— The business itself was run on gut feeling, with no clear view of what was working.

What I built

A single system with two connected sides.

1. Rafa — the AI assistant on WhatsApp

I named the assistant Rafa — personifying her was a deliberate design choice: a name makes a patient feel handled by someone, not interrogated by a form. She runs on a dedicated WhatsApp line, separate from the doctor's personal number, and:

— Answers incoming patients instantly, 24/7.

— Runs an initial triage to understand what the patient needs.

— Books evaluations and appointments on her own, with no input from the doctor.

New patients now get scheduled without the doctor ever touching her phone.

Rafa replying to a patient on WhatsApp in natural language, with personal data redacted.

Rafa handling a WhatsApp chat in natural language (patient data redacted).

2. The management dashboard

Everything the assistant handles flows into a panel where the doctor runs the clinic as a business, not just a calendar. It surfaces what drives decisions: goals vs. actuals (money in vs. out), average ticket, client retention, expenses, and the best-performing procedure. She went from "busy inbox" to actually seeing where the revenue comes from.

Clinic financial manager: monthly goal, revenue, expenses, profit, average ticket and charts by procedure.

Financial manager: goals, revenue, expenses and top procedure, in real time.

Where design made the difference

Code is easy to generate fast today. What no tool decides is what to build, for whom, and why one screen is better than another. That was my job — and it's the whole point of the project.

I drove every iteration: requesting features, cutting the ones that added noise, and pushing on the details that separate a demo from a product.

— Conversation design for Rafa — the flow of the triage, what she asks and in what order, so a patient feels handled rather than interrogated.

— What the dashboard shows, and in what hierarchy — the doctor doesn't need everything; she needs the few numbers that drive decisions, front and center.

— Legibility, spacing and contrast — deliberate typographic and layout choices so the panel is actually readable under time pressure.

— Responsive design — proper desktop and mobile layouts, because she checks the business from her phone.

— Performance — cache implementation so the panel loads fast and stays usable.

I don't write production code — I direct the build with design intent. That combination (design critique + fast execution) is exactly how the whole thing got shipped.

Architecture diagram: from patient on WhatsApp to triage, booking and management dashboard.

How the pieces connect — from patient to booking and management.

Outcome

— Patients no longer wait. Responses are near-instant, 24 hours a day.

— New clients book themselves. Scheduling no longer depends on the doctor being available.

— The doctor got her time back. The front-of-house bottleneck is gone, and her personal WhatsApp is hers again.

— The business is now legible. Retention, ticket, expenses and top procedures are visible at a glance instead of living in her head.

The stack

Supabase (data & backend) · Vercel (hosting) · GitHub (version control) · WhatsApp via the Meta Business API · Build tooling and the assistant's conversational logic.

What I took from it

A designer's real leverage today isn't pixels in isolation — it's shipping a product that solves a specific person's problem. I took a messy, personal-WhatsApp workflow and turned it into a system a clinic runs on every day, by pairing design judgment with fast execution.

That's the way I work: start from the real problem, decide what good looks like, and build it fast.

Clínica BC — an integrated AI + management system