DeepMind has unveiled its AI Co-Clinician research initiative, aimed at integrating multimodal AI agents into medical decision-making processes under physician supervision. Announced on April 30, 2026, the program explores the use of AI to assist in collecting patient histories, guiding physical examinations, and performing preliminary diagnostic reasoning. The system features real-time interaction capabilities via video and audio, emphasizing its role as a supportive tool rather than a replacement for clinical judgment. The AI Co-Clinician employs a dual-agent architecture and the NOHARM safety framework to ensure safe operation. In tests, the system achieved zero critical errors in 97 out of 98 primary care queries and matched or exceeded physician performance in 68 out of 140 evaluation dimensions, particularly in triage. The initiative is in collaboration with Harvard Medical School and Stanford School of Medicine, with plans for expanded clinical testing. DeepMind stresses the AI's role as a collaborative member in healthcare teams.