Intelligent virtual assistant for emergency care

Design and deployment of a multi-tenant generative artificial intelligence platform for a national medical scientific society in the field of emergency and urgent care.

The solution integrates a secure data lake unifying the institution's scientific evidence and documentary assets, until then scattered across different repositories and formats.

The analytical core implements a self-managed retrieval-augmented generation pipeline for the contextualised retrieval of medical knowledge in real time, combined with automated ingestion and crawling flows from external scientific databases.

A closed environment was designed with role-based access, qualification of administrator users and strict privacy controls, to guarantee the governance of institutional medical data. It combines Clinical processes and governed AI and data science.