Microsoft Corp. today announced the launch of Microsoft Discovery, a new enterprise-grade artificial intelligence platform built to revolutionize the research and development process by embedding “agentic AI” at every stage of scientific discovery.

Announced at the company’s annual Build 2025 conference this week, Microsoft Discovery is designed for scientists and engineers and uses a graph-based knowledge engine along with customizable AI agents to accelerate hypothesis generation, simulation and iterative experimentation.

Microsoft Discovery differs from traditional AI tools that focus on isolated tasks with a team-based model where specialized agents collaborate in real time across complex scientific workflows. The agents are tailored to specific domains, such as molecular simulation or literature review and work together under the orchestration of a central Copilot assistant. The setup allows researchers to manage end-to-end R&D cycles without needing deep computational expertise.

The platform’s graph-based knowledge engine is said by Microsoft to distinguish the offering from standard large language models by mapping nuanced relationships between proprietary and external scientific data. The core concept is that contextual reasoning assists researchers in navigating conflicting theories, diverse experimental results and assumptions with full transparency. Additionally, each step in the process is source-tracked to ensure traceability and trust in every AI-generated output.

Microsoft Discovery incorporates enterprise-grade governance, compliance and security controls to ensure responsible innovation. The service also allows researchers to extend the platform by integrating their own models, datasets and tools alongside Microsoft’s offerings and partner solutions.

Though formally announced only today at Build, Microsoft has already been testing Discovery and it says the early results are positive. In one case, the platform discovered a new coolant prototype for data centers in just 200 hours, an effort Microsoft claims would have traditionally taken years. The coolant was synthesized and validated in under four months, demonstrating the potential of AI-driven simulation and prediction in material science. Even before formal launch, Discovery has already uncovered a breakthrough that helps replace environmentally harmful “forever chemicals” in industrial applications.

With clear potential, unsurprisingly, Microsoft already has users lined up to use Discovery. Among them are British pharmaceutical and biotechnology GSK plc, who plans to leverage the service for parallel prediction and testing in drug development. Another early user, cosmetics company Estée Lauder Inc., plans to integrate Discovery into its innovation pipeline for personalized skincare and cosmetics.

Microsoft also plans to integrate Discovery with Nvidia ALCHEMI and BioNeMo NIM microservices to accelerate breakthroughs in materials science and life sciences. The integration will give researchers access to state-of-the-art tools for candidate identification, property mapping and AI model development for drug discovery. Running on Nvidia-accelerated Azure AI infrastructure, the collaboration aims to deliver faster, more scalable scientific outcomes across industries.

“Our goal is to bring the power of AI to scientists and engineers to transform the entire discovery process, from advanced knowledge reasoning and hypothesis formulation to experimental simulation and iterative learning,” said Aseem Datar, vice president of product innovation at Microsoft. “Microsoft Discovery enables researchers to collaborate with a team of specialized AI agents combined with a graph-based knowledge engine, to drive scientific outcomes with speed, scale and accuracy.”

Image: SiliconANGLE/Reve