FSP Anamnese Agent
We built an AI patient with 150+ cases and 7 personalities that doctors train with by voice
AI-powered virtual patient that simulates realistic history-taking dialogues for FSP Teil 1 preparation.
The build tracks the details candidates actually feel on exam day: duration, coverage, scoring depth, and language level.
Structured practice with timing, scoring, and review built into the session.
Structured practice with timing, scoring, and review built into the session.
Structured practice with timing, scoring, and review built into the session.
International medical graduates preparing for Germany's Fachsprachprüfung (FSP) had no realistic way to practice Teil 1 — the 20-minute patient interview. Sessions with native speakers were expensive and hard to schedule. Candidates needed to master 72 scoring criteria under time pressure, but had no system that could evaluate medical terminology, conversation structure, and completeness in real time.
We built Aida as a voice-first AI patient for MedSpeak AI's exam platform. Aida draws from 150+ realistic patient cases across all medical specialties, and responds through 7 distinct patient personalities — from cooperative to anxious to aggressive. Each 20-minute session is scored against 72 criteria in real time, with a competency radar tracking progress across sessions. The system includes a guided mode for beginners, free training mode for advanced users, and a 12-week structured study plan.
“Aida changed how our students prepare. They practice anytime, get instant feedback, and walk into the exam with real confidence.”
Next build
We design and ship production agents with the same focus: clear scope, measurable performance, and a usable operator surface.