AI avatars of health experts
Training healthcare professionals with avatars of recognised practitioners
Project
For a healthcare start-up, a continuing education platform was built around AI avatars of recognised experts. Each avatar is built with the expert, from their publications, courses and anonymised teaching cases. It reasons in the first person, cites the expert's documents in every answer and says what it does not know. Practitioners question it in writing or by voice, between two training sessions, and training organisations can integrate it into their courses. The platform is hosted in France, with Scaleway.
The need
A recognised expert trains a few dozen practitioners a year, at conferences or in university diploma courses. Between two sessions, their reasoning stays in their notes and courses, out of the learners' reach.
How it works
The expert uploads their corpus, then reviews how their avatar reasons. Nothing opens before their approval. Answers rely first on the expert's work, with sources. The platform asks for no identifying patient data and remains a training tool, never a clinical decision aid.
What the project shows
A university diploma course has used it since September 2026. Within a month, every learner surveyed wanted to keep their trainer's avatar, with twelve questions on average per active learner. In healthcare, trust rests on traceability: every answer points back to the expert's work.