Microsoft's MAI-DxO Multi-Agent Diagnostic System Scores 83% on Complex Cases

On October 23, 2026, TechCrunch and DEV Community both reported on a Microsoft AI diagnostic system. According to TechCrunch, Microsoft's Mustafa Suleyman said the system outperformed generalist doctors on difficult cases, with the research published in Science. The same report noted that most of the improvement appeared to come from the underlying AI models rather than from Microsoft's system itself.

DEV Community offered more detail. The system, called MAI-DxO, coordinates multiple AI agents to simulate a medical panel for complex diagnoses. The agents can question patients, order tests, and discuss cases with one another. On difficult cases drawn from a well-known medical journal, the system answered 83% correctly, compared with 36% for generalist and internal medicine physicians working with literature alone. The report also said the system was tested across models from several providers and achieved high accuracy on medical cases by orchestrating multi-agent reasoning workflows.

Both reports qualified the comparison. DEV Community noted that such cases are usually handled by specialist teams in real settings, and that specialist teams were not part of the comparison. TechCrunch attributed most of the gain to the underlying models rather than to Microsoft's system design. The available information stops at the results of the study published in Science; the materials contain no further developments.