Microsoft Build 2025 Advancing AI Agents, Developer Platforms, and the Agentic Web

/ AI, Application Development, Infrastructure
Microsoft Build 2025 Advancing AI Agents, Developer Platforms, and the Agentic Web

The News

At Microsoft Build 2025, Microsoft introduced major updates across its AI ecosystem, focusing on developer productivity, AI agent orchestration, and the infrastructure to support the emerging agentic web. Key announcements include the evolution of GitHub Copilot into a coding agent, Windows AI Foundry for on-device model deployment, Azure AI Foundry enhancements, and new standards like NLWeb and expanded Model Context Protocol (MCP) support. To learn more, visit the press release here.

Analysis

Microsoft is transforming its AI offerings from productivity features into a full-stack agentic AI ecosystem, supporting everything from LLM training to enterprise orchestration. By investing in both protocols (MCP, NLWeb) and platform-level tooling (Foundry, Copilot Studio, Azure AI), Microsoft is aiming to standardize how AI agents are built, governed, and deployed. For developers, enterprises, and research organizations, this provides a scalable, secure, and open foundation for building the next generation of intelligent applications.

GitHub Copilot Becomes an Asynchronous Coding Agent

GitHub Copilot is evolving from an IDE assistant into a fully agentic AI platform, with new asynchronous capabilities integrated across the GitHub platform. Developers can now delegate complex workflows like PR management, documentation generation, and environment setup to Copilot’s agent, transforming productivity for software teams. This shift places GitHub Copilot in direct competition with developer-centric agent platforms emerging in the AI tooling market.

Windows AI Foundry & Foundry Local: LLMs at the Edge

Microsoft is doubling down on edge AI deployment with Windows AI Foundry and Foundry Local, enabling developers to manage, fine-tune, and run open source LLMs across local devices or proprietary models via simplified APIs. Foundry Local ensures compatibility across client and cloud without vendor lock-in, enabling use cases like offline summarization, image-to-text, or secure document inference in air-gapped environments.

Azure AI Foundry Adds xAI’s Grok Models and New Tooling

Azure AI Foundry continues to serve as Microsoft’s unified development platform for agentic AI. Highlights include:

  • Grok 3 and Grok 3 Mini LLMs (from xAI) available in Azure
  • Model Leaderboard for comparing model quality across tasks
  • Model Router for dynamically selecting models based on workload

These additions expand Azure’s LLM ecosystem, empowering developers to rapidly evaluate and deploy models with confidence in real-world scenarios.

Advancing Secure, Scalable AI Agents

Azure AI Foundry Agent Service & Observability

New features include:

  • Support for multi-agent workflows
  • Open protocol compatibility (A2A, MCP)
  • First-party tool integrations (e.g., SharePoint)
  • Built-in observability for cost, quality, safety, and latency metrics

These advancements are designed to operationalize AI agents across enterprise environments, ensuring scalability and governance.

Microsoft Entra Agent ID

With Microsoft Entra Agent ID, every agent created in Copilot Studio or Azure AI Foundry receives a unique identity within Entra’s directory services. This solves the growing challenge of “agent sprawl” by enabling lifecycle and policy management of AI agents from day one.

Copilot Tuning and Multi-Agent Orchestration

  • Copilot Tuning allows customers to fine-tune models with proprietary data inside Microsoft 365 boundaries—ideal for regulated industries.
  • Multi-agent orchestration in Copilot Studio supports dynamic task sharing between agents, unlocking complex enterprise workflows and domain-specific capabilities.

Building the Open Agentic Web

NLWeb: HTML for Natural Language Interfaces

Microsoft introduced NLWeb, a new open protocol designed to bring semantic interaction to the web. With just a few lines of code, websites can expose content to natural language agents, enabling multimodal experiences without bespoke integrations.

First-Party Support for MCP

Microsoft announced broad support for the Model Context Protocol (MCP) across its products, including GitHub, Copilot Studio, Azure AI Foundry, Semantic Kernel, and Windows 11. Microsoft joined the MCP Steering Committee and made two open contributions to the ecosystem, helping standardize contextual alignment for LLMs and agents.

Microsoft Discovery: AI for Scientific Innovation

Microsoft unveiled Microsoft Discovery, a research-focused platform designed to revolutionize the scientific process using agentic AI. From hypothesis generation to simulation and experimental analysis, Discovery enables faster, more scalable R&D in pharmaceuticals, materials science, and other domains.

Author

  • Paul Nashawaty

    Paul Nashawaty, Practice Leader and Lead Principal Analyst, specializes in application modernization across build, release and operations. With a wealth of expertise in digital transformation initiatives spanning front-end and back-end systems, he also possesses comprehensive knowledge of the underlying infrastructure ecosystem crucial for supporting modernization endeavors. With over 25 years of experience, Paul has a proven track record in implementing effective go-to-market strategies, including the identification of new market channels, the growth and cultivation of partner ecosystems, and the successful execution of strategic plans resulting in positive business outcomes for his clients.

    View all posts