Medical AI, Natural Language Processing, Persona Agents, User Simulation
Run a multi‑turn interrogation to score agent consistency