AI systems capable of clinical reasoning and dialogue can increase access to medical expertise. The Articulate Medical Intelligence Explorer (AMIE) previously demonstrated expert-level text-based diagnostic dialogue. Now, in “Towards expert-level medical AI for real-time video consultations”, Google presents AMIE (Video), a real-time video configuration built on Gemini and Project Astra.

AMIE (Video) uses an asynchronous multi-agent architecture with three specialized agents: Talker (patient-facing, low-latency interaction), Planner (background clinical reasoning), and Perception (audio-visual cue analysis). This design balances conversational speed with deep reasoning. An automated evaluation suite based on a taxonomy of clinical audio-visual competencies guided development.

In a large-scale randomized OSCE study with 100 scenarios, 300 consultations, and 30 board-certified primary care physicians (PCPs), AMIE (Video) was evaluated against a text-only AMIE and PCPs via video. Clinical evaluators rated AMIE (Video) on par with PCPs across core competencies like history-taking, diagnostic accuracy, and communication. This is the first demonstration of an AI system exhibiting expert-level performance in real-time clinical video consultations.