Microsoft Research

Microsoft Research conducts fundamental research in computing, AI, and science, sharing models, datasets, and software that researchers and developers can build upon. Its Orchard framework lets researchers train and evaluate AI agents across different tasks using shared infrastructure. Aurora, its foundation model for weather and Earth systems, supports weather, climate, and energy applications, including probabilistic ensemble forecasting. These tools sit within a broader research program spanning scientific discovery, human health, and computing.

Founded by Richard (Rick) Rashid in 1991, the organization is now led by executive vice president Igor Tsyganskiy, who succeeded Peter Lee in 2026 as Lee became president of Microsoft Science. Its engineering contributions include AutoGen, a framework for applications in which multiple AI agents act autonomously or work alongside people. Its architecture separates message passing and agent runtimes from higher-level conversation APIs. AutoGen remains available in community-managed maintenance mode; its repository directs new projects to Microsoft Agent Framework.

Microsoft Research combines academic-style exploration with an industrial role: its researchers collaborate with Microsoft product teams to shape roadmaps and product innovation, alongside partners in academia, government, and industry. As of 2026, the organization reported that its community and collaborators had published tens of thousands of papers and released hundreds of open-source projects and datasets over the preceding three decades, making shared research a substantial part of its output.