FLORENCE
AI medical secretary on WhatsApp using natural language processing. The project prioritized faster patient service, a better user journey, and scalable operations with strong governance.
InfectoX faced operational bottlenecks in patient care: slow responses to patient inquiries, lack of automation in repetitive processes, and difficulty scaling support without increasing costs. There was an opportunity to implement an intelligent virtual secretary on WhatsApp to improve user experience and optimize internal resources.
I developed an orchestration architecture with N8N as the automation core, integrating unofficial WhatsApp APIs (Evolution and WAHA) for real-time communication. The solution was containerized with Docker and deployed in self-hosting infrastructure via Coolify, ensuring full control, data security, and scalability. The MVP includes natural language processing, contextualized automatic responses, and intelligent routing of inquiries to specialists when necessary.
I served as Tech Lead and Product Owner: architecture definition, hands-on learning in N8N and workflow orchestration, WhatsApp API integration, container and infrastructure setup, security and data governance, technical-commercial alignment with stakeholders, and MVP delivery in production environment.
Functional and validated MVP in a real environment. 70% reduction in response time for common inquiries. Rapid prototyping of AI solutions without dependency on proprietary APIs. Scalable and independent infrastructure. Project currently in commercial feasibility study phase with clear roadmap for feature expansion and integration with clinic legacy systems.
Unofficial WhatsApp APIs offer flexibility but require constant monitoring; N8N is highly productive for complex orchesturation in short time; self-hosting with Coolify offers perfect balance between control and operability; continuous validation with real users is critical before scaling; internal education about AI capabilities is fundamental to maximize ROI.