Speed up Research with Microsoft Discovery
Ali Zaman explains Microsoft Discovery and how it coordinates specialized AI agents to help scientists run an evidence-driven research loop, using battery design as the example and focusing on why search and summarization aren’t enough for real R&D workflows.
Overview
Microsoft Discovery is presented as an agentic research and development (R&D) system that coordinates multiple specialized AI agents to:
- Analyze research evidence
- Propose candidate materials
- Run scientific tools
- Validate results
A key point is that the scientist remains in control: they set goals and make the key decisions, while the system iterates continuously as new evidence arrives.
Why R&D needs more than search and summarization
The video frames the problem as research evidence being spread across multiple sources and modalities, including:
- Research papers
- Proprietary data
- Simulations
- Lab experiments
In that context, simple search and summarization are described as insufficient for driving decisions about what to test next.
How Microsoft Discovery works (continuous research loop)
Discovery is described as operating as a continuous loop where:
- The plan shifts toward the most promising candidates as new evidence arrives
- Specialized agents coordinate to move from evidence to candidate proposals to tool execution and validation
- The workflow mirrors the scientific method, aiming to reach results with fewer experiments
Example used in the video: battery design
Ali Zaman uses battery design as the running example, with the target outcome described as a battery that:
- Charges faster
- Lasts longer
- Uses safer materials
Resources
- Microsoft Discovery: https://azure.microsoft.com/en-us/solutions/discovery
- Blog: Microsoft Discovery: Advancing agentic R&D at scale: https://azure.microsoft.com/en-us/blog/microsoft-discovery-advancing-agentic-rd-at-scale/
- What is Microsoft Discovery?: https://learn.microsoft.com/en-us/azure/microsoft-discovery/overview-what-is-microsoft-discovery
- Explore more Azure resources!: https://azure.com/AzureEssentials
Video chapters
- 0:00 Why building a better battery takes more than search
- 0:56 Why the battery research space grows so fast
- 1:14 How Microsoft Discovery coordinates AI agents for research
- 1:35 Why scientists stay in control of AI-driven research
- 1:50 How Microsoft Discovery works as a continuous research loop