Microsoft’s Vision for the Open Agentic Web: How AI Agents Are Transforming Enterprise Computing Systems and Research
At the Microsoft Build 2025 conference, Microsoft presented its vision of the future, centered around the so-called “open agentic web” and a new era of autonomous AI agents. This vision represents a fundamental shift in the way people interact with technology and changes the paradigm of the internet itself. As Microsoft Chairman and CEO Satya Nadella said, we are entering a new era of computing in which AI agents function as digital assistants that can understand our intentions, delegate tasks among themselves, and act autonomously on our behalf.
Microsoft’s Vision
The era of agentic AI brings a fundamental change in how organizations approach software development and process automation. Microsoft believes the future belongs to a world where AI agents are not merely tools, but active participants in the digital environment. These agents are designed to overcome the limitations of traditional chatbots—they can remember, plan, learn from previous interactions, and collaborate with other agents to solve complex tasks. Agents are not merely reactive, but genuinely proactive, capable of anticipating users’ needs and working autonomously to fulfill them.
At the Build 2025 conference, Microsoft introduced several key innovations in multi-agent orchestration. The new multi-agent orchestration feature in Copilot Studio, now available in preview, enables organizations to create systems in which multiple specialized agents can collaborate to achieve complex goals. These agents—created using tools such as Microsoft 365 Agent Builder, Azure AI Agents Service, and Azure Fabric—can now delegate tasks among themselves. In practice, this means that one agent can, for example, retrieve sales data from a CRM system, pass it to another agent to prepare a proposal in Word, and then initiate the planning of follow-up steps through Outlook. This approach dramatically simplifies the automation of complex business processes and opens up new opportunities to increase productivity.
Microsoft Discovery, a new platform introduced at Build 2025 and focused on transforming research and development through agentic AI, plays a significant role in this transformation. Microsoft Discovery is designed for scientific and research teams that need to process and analyze vast amounts of specialized data and publications. As Kevin Scott, Microsoft’s Chief Technology Officer, explained, Microsoft Discovery uses advanced agent technologies to help researchers better navigate the flood of scientific literature, gain new insights, and accelerate scientific discoveries. The platform is built on three key pillars: agents specialized in scientific fields, knowledge graphs for organizing information, and advanced search that enables precise querying across enormous corpora of scientific texts.
For this vision to work in practice, Microsoft is working intensively to create open protocols and standards that ensure interoperability among agents, applications, and services. The Model Context Protocol (MCP) plays a key role, described by Microsoft as the “universal USB-C connector for AI.” MCP enables communication and the exchange of context between different types of agents and tools, such as GitHub Copilot Studio and Azure Foundry. Another significant project is the Natural Language Web (NLWeb), led by R.V. Guha at Microsoft. The goal of this project is to enable websites to communicate natively through natural language, allowing them to be transformed into “agentic applications” that interact directly with users’ requests through conversation.
As AI agents become increasingly autonomous, the need for security and governance is also growing. Microsoft introduced Microsoft Entra Agent ID as part of an expansion of Zero Trust principles specifically for identity management—not only for people, but also for every deployed agent. Each new agent receives its own unique identity, centrally managed through Entra directory services, enabling robust authentication and authorization while simplifying management at scale. Integration with compliance solutions such as Microsoft Purview ensures that even highly regulated industries can adopt these technologies securely.
Large-Scale Deployment of Agents in Businesses
Adoption statistics show rapid enterprise uptake—more than 230,000 organizations, including over 90% of Fortune 500 companies, are already building their own automations using Copilot Studio. The enhanced “agent mode” of GitHub Copilot now serves more than 15 million developers worldwide. These figures underscore how quickly generative and agentic AI is becoming foundational across various industries—from legal automation to accelerating scientific research.
Microsoft Discovery demonstrates a concrete application of agentic AI in research and development. The platform enables the creation of specialized agents that can search, analyze, and synthesize insights from extensive databases of scientific publications. Developers and researchers can use Microsoft Discovery to discover knowledge more quickly and innovate more efficiently. Microsoft cites examples of use in the pharmaceutical industry, where agents help accelerate drug development, and in materials research, where they enable the faster identification of new materials with desired properties. A unique feature of Microsoft Discovery is also the ability to connect agents with laboratory instruments and automated experiments, further accelerating the scientific discovery cycle.
Microsoft Is Taking the Open-Source Path
Microsoft’s commitment to openness is also evident in the fact that GitHub Copilot Chat was opened as open-source for Visual Studio Code. Developer SDKs are being unified, and an increasing number of core tools are being provided as open-source. Windows AI Foundry offers APIs supporting both proprietary and open-source models for vision- and language-related tasks. This approach reduces the risk of vendor lock-in and supports innovation throughout the ecosystem.
The era of agentic AI brings both opportunities and challenges. Key benefits include democratizing access to AI through low-code/no-code development tools, open standards that reduce the risk of dependence on a single vendor, and enterprise security features addressing regulatory concerns. At the same time, there are risks associated with privacy management when decision-making authority is delegated, as well as with ensuring transparency and accountability when autonomous systems act on behalf of individuals or businesses.
According to Microsoft, the future of agentic AI is not merely about improving existing tools, but about fundamentally changing how we interact with the digital world. AI agents will become increasingly autonomous, capable, and integrated into our daily lives and work processes. As Microsoft continues to develop its vision of the open agentic web and platforms such as Microsoft Discovery, we can expect further breakthroughs in how AI transforms businesses, research, and everyday productivity. The Microsoft Build 2025 conference demonstrated that the era of AI agents is not a distant future, but is rapidly becoming a reality.
