Past project
ProntoVA
ProntoVA was a WhatsApp receptionist platform for appointment-based clinics: one backend running many bots, each defined by a config file. I built it end to end, alone, and stopped it on 2026-09-03. It is not running, and it never had a customer.
My role
Solo. The product, the code, and the GDPR work. No co-founder, no client brief, no agency.
The decision
I chose a narrow buyer instead of a general-purpose chatbot framework: appointment-based clinics in Madrid — dental, aesthetic, physio. They take inquiries on WhatsApp all day, and because they are health-adjacent, GDPR is the differentiator rather than a tax.
The channel decision is the one worth reading. V1 ran on Baileys, an unofficial WhatsApp library, and it worked: a real five-turn conversation through the full pipeline on 2026-07-07. Deployed to a datacentre it failed deterministically — WhatsApp accepted the pairing, then killed the session with a 401 five seconds later. That is IP policy, not a bug. A residential workaround existed; I did not take it, because the bar is a channel a clinic can depend on. On 2026-07-08 I made the official WhatsApp Cloud API the pilot channel and traded my launch date for it.
What I built
Channel, model, payment and speech-to-text are each swappable per project in config, which is why changing channel meant a new adapter, not a rewrite. Tenancy is enforced by isolation tests, every model call is metered to a cost, and a privacy layer — retention purge, erasure, opt-out, PII redaction in logs — was built and reviewed.
One review catch is a product judgment, not a code bug: my opt-out keyword list originally matched the bare Spanish words cancelar and baja. In a dental clinic those mean "cancel my appointment" and "sick note" — a patient typing either would have been silently and permanently blocked. The list was narrowed to unambiguous unsubscribe phrases, and a test now asserts those two inputs must not match.
Code is private; walkthrough available on request.
What it shows I can build a whole AI product, and I stop one on purpose when it can't go forward.
Why I stopped
I built ProntoVA, a WhatsApp + LLM booking assistant for clinics, end to end. Stopped before launch pending a data-protection review. It is no longer running. Nobody was contacted, nothing was sold, and the product site has been taken down. The platform was built and exercised, never launched, and I do not claim users or revenue.
Stack
TypeScript on Node, Anthropic API behind a swappable model adapter, BullMQ on Redis for durable per-conversation queuing, Postgres with pgvector enabled on day one (RAG not built), Docker Compose, Caddy, zod, vitest.