Microsoft Unveils AI Agent Network Vision, 1.3 Billion Agents by 2028

Word on the StreetMonday, May 19, 2025 1:03 pm ET
2min read

Microsoft has unveiled an ambitious vision for the future at its annual Build conference in Seattle, where it outlined a future where users worldwide will complete tasks ranging from coding to operating Windows systems through AI agents. The company envisions an "open agent network" where AI agents can make decisions and execute tasks for individual users or entire organizations.

AI agents, a significant trend in the tech industry, are semi-autonomous or fully autonomous AI software capable of performing specific tasks for users, such as migrating data between different applications or booking concert tickets. Some agents can even communicate with each other, forming a network capable of completing more complex tasks.

Scott Guthrie, Microsoft's executive vice president of cloud and AI, highlighted the accelerating development of AI, noting the shift from conceptual validation to practical solutions that drive business impact. He emphasized that this trend is expected to continue and accelerate, particularly with the rise of what Microsoft calls the "agent network." The company's key goal is to make it easier for enterprises, developers, and startups to keep pace with all the new technologies.

Microsoft revealed that approximately 230,000 organizations are already using its Copilot Studio to develop custom AI agents. The company projects that by 2028, enterprises will launch 1.3 billion agents. During the conference, Microsoft showcased several AI agent applications, including the Microsoft 365 Agents Toolkit for developers to build agents for Microsoft 365, the GitHub Copilot coding agent to assist developers in writing code, and the direct integration of AI agents with native Windows applications.

ODP Corporation, the parent company of office supply brand Office Depot, is leveraging Microsoft's technology to develop AI agents aimed at enhancing customer outreach. For instance, the agents use customer purchase history, product pricing, and other relevant data to suggest when sales representatives should contact customers for sales opportunities. This approach minimizes the barriers to information retrieval and effectively plans the daily schedule for sales teams, prioritizing tasks based on customer importance and satisfaction levels.

Microsoft is also focusing on enabling customers to create "multi-agent systems" using its Copilot Studio application, a low-code tool that allows users to quickly build their own AI agents. Through multi-agent systems, Microsoft's customers will be able to program agents to communicate with each other and execute multiple tasks. For example, a Copilot Studio agent could extract sales data from a customer relationship management application, pass it to a Microsoft 365 agent to draft a proposal in Word, and then trigger another agent to schedule a follow-up in Outlook.

All of this is happening as Microsoft continues to invest tens of billions of dollars in the AI sector, with Wall Street closely monitoring the returns on these investments. The company plans to spend approximately 80 billion dollars to expand its AI data centers and train and deploy AI models across various business units. However, the software giant acknowledges that it faces resource constraints, as the number of users attempting to access AI features has exceeded Microsoft's processing capacity, potentially slowing the adoption and sales of these applications.

In its recent financial report, Microsoft indicated that AI contributed 16 percentage points to the growth of its Azure business, a strong signal of the effectiveness of its investments. However, the company also reported that it expects resource limitation issues to persist beyond the June quarter.