Hyderabad: Microsoft Build 2025 was all about agentic artificial intelligence as the software major announced a broad set of advancements in AI agents across GitHub, Azure AI Foundry, Windows, and Microsoft 365, promoting an "agentic web" where intelligent systems perform tasks autonomously.
Microsoft envisions the internet as an open, agentic web, where AI agents make decisions and perform tasks on behalf of users or organisations. The Microsoft Build 2025 was basically a showcase of how the Windows maker is trying to make this vision a reality through its platforms, products, and infrastructure.
AI for software development
Microsoft is introducing new features across GitHub, Azure AI Foundry, and Windows to help developers work faster and scale their projects. First of all, GitHub Copilot is evolving from an in-editor assistant to an AI coding agent, supporting asynchronous coding within GitHub. The Azure AI Foundry, a unified platform for developers to design, customise, and manage AI agents and applications, now includes Grok 3 models from xAI, providing access to 1,900+ AI models to developers.
More capable and secure AI agents
At Build 2025, Microsoft unveiled new pre-built agents, custom agent building blocks, multi-agent capabilities, and new models to help developers and organisations build and deploy agents securely. The Azure AI Foundry Agent Service allows developers to orchestrate multiple specialised agents to handle complex tasks, integrating Semantic Kernel and AutoGen into a single developer SDK. It also introduces Agent-to-Agent (A2A) and Model Context Protocol (MCP) support.
Microsoft Entra Agent ID (in preview) allows for secure identity management for AI agents, reducing "agent sprawl" and enhancing governance with Purview data security and compliance controls. Microsoft 365 Copilot Tuning enables businesses to train AI models and create agents using their proprietary data for domain-specific tasks, improving efficiency within the Microsoft 365 service boundary. Additionally, multi-agent orchestration in Copilot Studio allows agents to collaborate on complex tasks.
Microsoft Discovery to accelerate scientific research
At Build 2025, the software major introduced Microsoft Discovery, a new AI-powered platform designed to accelerate scientific research and development. By leveraging agentic AI, the platform helps researchers streamline the discovery process across industries, enabling faster innovation and reducing time to market for new products.
Supporting the open agentic web
Microsoft is advancing open standards and infrastructure to support AI agents more effectively. Key initiatives include Model Context Protocol (MCP) Expansion across its agent platforms and frameworks across GitHub, Copilot Studio, Dynamics 365, Azure AI Foundry, Semantic Kernel, and Windows 11. Microsoft and GitHub are also contributing to MCP’s development by enhancing authorisation methods and creating an MCP server registry for managing centralised repositories.
It also announced the NLWeb Project, a new open framework designed to function like HTML for AI-powered websites, enabling conversational interfaces where users interact with content using models of their choice. Every NLWeb endpoint is also an MCP server, so websites can make their content easily discoverable and accessible to AI agents if they choose, Microsoft explained.
