Mozt Clinics
WhatsApp-native AI agent for clinic appointment booking in LatAm
Context
Small and medium clinics in Guatemala lose a meaningful share of inbound patient inquiries because the front desk can't be available 24/7. WhatsApp is the de facto channel for patient communication in the region, but small clinics don't have the operational scale to staff it around the clock or build their own automation.
What I did
Co-founded Mozt with three peers. As CTO and product lead, I architected the system, made the core stack decisions, and led the technical build of the agent layer and integrations.
How it works
A WhatsApp-native conversational agent powered by Claude Haiku handles inbound patient messages, qualifies the consultation type, and books directly into the clinic's Google Calendar. The clinic's team gets a dashboard with daily leads, upcoming appointments, and conversation history. Built on Next.js + Supabase + Vercel, with Twilio as the WhatsApp BSP.
A non-obvious decision: we initially planned to integrate directly with Meta's WhatsApp Business API, but verification delays were blocking weeks of iteration with real prospects. We switched to Twilio. Higher per-message cost, but we recovered the velocity to keep shipping. Speed over margin at this stage.
Outcomes
Currently in pilot phase. Demo video and onboarding flow are the immediate priorities. The deeper bet is that small-to-medium clinics in LatAm are underserved by both the WhatsApp Business ecosystem and by horizontal SaaS tools — and that a specialized vertical product can win this niche.
What I'm intentionally not sharing publicly: pricing, specific clinic names, prompt engineering details, and parts of the roadmap. Reach out if you want to talk seriously about the product.