Microsoft Unveils Agent 365: Managing AI Workers Like Employees

In a pivotal shift for corporate computing, Microsoft is unveiling a suite of tools that treats AI agents as a digital workforce requiring management similar to human employees. At this week’s Ignite conference in San Francisco, the tech giant is introducing Microsoft Agent 365, positioning it as a management system for the growing legion of autonomous AI assistants working alongside humans in modern businesses. This marks a significant step toward Satya Nadella’s vision of a future where Microsoft’s business expands beyond per-user revenue to encompass “per agent” income streams as well, a concept he recently elaborated on during an appearance on Dwarkesh Patel’s podcast. As businesses increasingly deploy AI agents to handle everything from sales outreach to document creation, Microsoft is betting that organizations will need robust systems to oversee these digital workers, much as HR departments manage human staff.

The centerpiece of Microsoft’s announcement, Agent 365, functions as a “control plane” within the familiar Microsoft 365 Admin Center that IT teams already use. Its primary purpose is revolutionary yet intuitive: to govern an organization’s entire AI workforce by assigning each agent a unique identification. This approach allows companies to apply their existing security infrastructure to track agent activities, control data access permissions, and prevent security breaches or data leaks. By building this system into tools that businesses already use—such as Microsoft Entra (formerly Active Directory)—the company is making adoption relatively seamless for organizations already embedded in the Microsoft ecosystem. This addresses the growing “Shadow AI” problem that has plagued businesses throughout 2025, where employees independently adopt unmanaged AI tools outside company oversight.

Microsoft isn’t alone in recognizing this opportunity. The race to dominate the AI agent management space is intensifying, with Google’s Gemini Enterprise platform, Amazon’s Bedrock AgentCore tools, and Salesforce’s Agentforce 360 all vying for market share. The stakes are enormous—analysts project the AI agent market to grow from approximately $7.8 billion in 2025 to over $50 billion by 2030, with an estimated 1.3 billion agents expected to be in operation by 2028. These figures help explain why tech giants are investing so heavily in agent infrastructure, seeing it as a crucial revenue stream to offset their massive investments in AI development and the computing power needed to run these systems. Microsoft’s approach leverages its entrenched position in corporate identity and security systems, giving it a potential advantage in this high-growth market.

Beyond the management framework, Microsoft is showcasing several specific agent implementations that demonstrate the practical applications of this technology. Perhaps most striking is the “fully autonomous” Sales Development Agent, which can independently research, qualify, and engage sales leads—essentially functioning as a digital sales team member. In the security realm, Security Copilot agents in Microsoft’s tools will help IT teams automate tasks like creating new security policies from simple text prompts. The company is also enhancing its productivity suite with agents for Word, Excel, and PowerPoint that can create complete, high-quality documents or presentations from scratch based on natural language requests. These implementations illustrate how AI agents are moving beyond simple assistants to become specialized digital workers handling complex tasks with minimal human oversight.

The technical architecture behind these advancements reveals Microsoft’s strategic thinking. The new “Agent Workspace” in Windows creates a secure, isolated environment where agents can execute complex tasks using their own distinct identities, allowing IT departments to monitor their activities. By giving each agent its own identity within Microsoft Entra—the same system that handles employee authentication and permissions—Microsoft is treating these digital entities much like human workers from a systems perspective. This integration means businesses can apply familiar security protocols and compliance measures to their AI workforce, reducing the need for specialized knowledge or separate management systems. Microsoft is offering Agent 365 in preview through its Frontier early-access program starting this week, though pricing details have not yet been announced.

As these technologies mature, they represent a fundamental shift in how organizations will structure their workforces. The line between human and AI labor is becoming increasingly blurred, with digital agents taking on roles that would previously have required human judgment and decision-making. Microsoft’s announcements suggest a future where corporate org charts include both carbon and silicon-based workers, managed through parallel systems that increasingly resemble each other. For businesses, this promises efficiency gains and cost savings, but also introduces new challenges around oversight, accountability, and integration. For Microsoft, it represents a significant expansion of its potential market—no longer limited by the number of human users, but now encompassing the potentially much larger population of AI agents that organizations might deploy to augment their human workforce. As Nadella suggested, the future of Microsoft’s business model increasingly depends not just on how many people use its software, but on how many digital agents do as well.