The tech giant reveals new tools and standards to power the next wave of AI agents across enterprise, development, and scientific domains.
At its annual Build developer conference on May 19, Microsoft unveiled a bold new vision for the future of artificial intelligence – one driven by autonomous, task-oriented agents that promise to redefine how applications are built, deployed, and managed.
Dubbed the “Agentic” future of AI, Microsoft’s strategy focuses on enabling intelligent systems that can act independently, collaborate with other agents, and dynamically adapt to user and enterprise needs. The company introduced a suite of tools, platforms, and protocols designed to support this shift, marking a significant evolution from smart models to self-directed AI agents.
AI Agents at the Core
At the heart of Microsoft’s AI roadmap is its expanding Copilot ecosystem and Azure AI Foundry, a unified platform for developing, managing, and deploying AI systems. The company revealed that over 230,000 organizations, including 90% of the Fortune 500, are already leveraging Copilot Studio to build AI agents and automate tasks. Meanwhile, GitHub Copilot – its AI-powered coding assistant – has surpassed 15 million users globally.
New features across Microsoft 365 and GitHub now support asynchronous coding agents, multi-agent orchestration, and domain-specific training, enabling developers to shift from traditional manual workflows to autonomous systems that enhance collaboration, productivity, and operational intelligence.
Empowering Developers and Enterprises
To support AI innovation at scale, Microsoft announced an updated Windows AI Foundry, providing developers with a consistent environment to train, fine-tune, and deploy AI models across local and cloud environments. Tools like the Model Leaderboard and Model Router will help teams identify the most effective models for specific tasks based on performance metrics.
Recognizing the rising importance of governance in autonomous systems, Microsoft introduced Microsoft Entra Agent ID (currently in preview), a new identity solution that assigns unique digital IDs to agents, allowing enterprises to securely monitor and manage agent behavior. Integration with Microsoft Purview will further enhance compliance through features like automated audits and detailed reporting.
The Azure AI Foundry Agent Service will now offer robust observability with real-time dashboards tracking agent performance, safety, cost-efficiency, and quality – giving enterprises clear visibility into the behavior of their AI deployments.
Pioneering the Agentic Web
To help define the architecture of the internet’s next phase, Microsoft announced support for the Model Context Protocol (MCP) and proposed a new open standard called NLWeb – a semantic web framework designed to serve content not only to human users, but also directly to AI agents. The initiative envisions NLWeb as the “HTML for the agentic web”, creating a more AI-native online ecosystem.
Accelerating Scientific Discovery
Microsoft also unveiled Discovery, a new platform designed for the scientific community. Powered by agentic AI, Discovery aims to accelerate innovation across fields such as drug development, climate science, and materials research by streamlining the research lifecycle through automation and intelligent coordination.
The Road Ahead
Throughout the Build 2025 conference, Microsoft underscored that developers are central to this AI transformation. With an emphasis on open-source tools, extensible infrastructure, and adherence to emerging industry protocols, the company is positioning itself as the foundational platform for building the next generation of AI systems.
As AI transitions from model-centric to agent-centric design, Microsoft’s announcements reflect a rapidly evolving software landscape—one in which intelligent agents not only assist, but actively build, debug, and deploy software in real time.
