Microsoft's annual Build developer conference, held in Seattle this week, showcased a significant push towards an "open agentic web," dominated by AI agents, enhanced Copilot functionalities, and intelligent discovery tools. CEO Satya Nadella underscored the company's commitment to reshaping the software development lifecycle with AI, making it easier to construct capable and secure AI agents, accelerate scientific discovery, and promote open standards.

A central theme was the evolution of AI agents. Microsoft is introducing tools and platforms designed to empower developers to build more sophisticated agents capable of interacting, making decisions, and performing tasks autonomously on behalf of individuals, teams, and entire organizations. Azure AI Foundry has been significantly enhanced, now supporting over 10,000 models, encompassing open-source, proprietary, and task-specific options. This allows developers to fine-tune models to meet specific business requirements, with expanded model offerings including Grok 3 and Grok 3 Mini.

GitHub Copilot is transitioning from a pair programmer to a "peer programmer," with the introduction of GitHub Coding Agent ("Project Padawan"). Generally available to Copilot Enterprise and Copilot Pro+ users, this agent can handle tasks of low-to-medium complexity in well-tested codebases, such as bug fixes, feature additions, code refactoring, and documentation improvements. It integrates context from discussions and adheres to repository instructions, while security measures ensure careful review of contributions. Furthermore, Microsoft plans to open source the GitHub Copilot Chat Extension for VSCode.

Microsoft 365 Copilot is gaining personalization capabilities through "Copilot Tuning," a low-code feature within Copilot Studio. This allows companies to customize AI models with their data, processes, and workflows, creating domain-specific AI agents without coding expertise. This means a legal firm, for example, can develop an agent that automates tasks by combining institutional knowledge with client-specific context. Copilot Studio is also gaining "pro" developer features, positioning it as a tool for professional developers needing powerful tools to build complex agents. Multi-agent systems can now be built in Copilot Studio, where agents delegate tasks to one another, constructed with the Microsoft 365 Agent Builder, Azure AI Agents Service, and Azure Fabric.

Microsoft is also addressing the need for a safer AI ecosystem. It's championing the Model Context Protocol (MCP) as a standard for managing AI agent interactions across environments. MCP support is being integrated across GitHub, Copilot Studio, Dynamics 365, Azure, Azure AI Foundry, Semantic Kernel, Foundry Agents, and Windows 11, envisioning it as a "universal USB-C connector for AI." A new open standard called NLWeb enables websites to communicate with AI in natural language, allowing users to interact with sites by asking questions instead of navigating menus.

To accelerate scientific breakthroughs, Microsoft unveiled Microsoft Discovery, an AI-powered platform designed to transform research and development processes. Discovery uses agentic AI to automate complex tasks, enabling scientists and engineers to accelerate innovation through streamlined data analysis and experimentation. Integrating with existing tools and datasets facilitates collaboration and knowledge sharing across teams. Microsoft researchers have already used Discovery to discover a novel coolant prototype for data centers in approximately 200 hours, a process that would typically take months or years.

Furthermore, Microsoft is embedding security features directly into Azure AI Foundry and Copilot Studio. Entra Agent ID, now in preview, manages agents built using Microsoft tools and integrates with third-party platforms like ServiceNow and Workday. The company also announced SQL Server 2025, now in public preview, touting the integration of AI directly into the database engine.

These advancements collectively represent Microsoft's strategy to deeply integrate AI across various platforms, encouraging companies to build tailored AI solutions while also supporting scientific breakthroughs using advanced AI technologies.