San Francisco. At its Ignite 2025 conference, Microsoft announced one of the biggest changes in business computing since cloud services became popular. Companies will now manage AI agents like they manage human employees.
Satya Nadella has hinted at this shift for months. In a recent interview on Dwarkesh Patel’s podcast, the Microsoft CEO suggested that the tech industry needs to think not only in terms of “per user” but also “per agent.” That idea is now taking shape in the company’s product strategy.
This week at Ignite, Microsoft introduced Agent 365, a new platform designed to standardize, secure, and supervise the growing number of autonomous AI agents in organizations. This system treats AI agents as key digital workers, each with unique identities, permissions, and monitoring systems, all governed by familiar enterprise controls.
A Control Plane for the AI Workforce
The heart of Microsoft’s announcement is Microsoft Agent 365, a control system integrated directly into the Microsoft 365 Admin Center. For IT teams managing thousands of employee accounts, devices, and applications through Microsoft 365, Agent 365 is a logical extension of their existing tasks.
The idea is straightforward: AI agents are becoming workers in their own right. They perform tasks, access data, create content, analyze systems, and make decisions that can affect business operations. But unlike human users, their actions are often unclear, isolated, or unmanaged. Microsoft wants to change that.
Agent 365 gives every AI agent a distinct digital identity similar to an employee ID badge. These identities are stored in Microsoft Entra, the cloud identity service previously known as Active Directory, where they receive the same access rules, permissions, and monitoring that govern human employees.
By extending identity management to AI systems, Microsoft aims to:
- Prevent unauthorized data access
- Reduce the risk of AI-driven data leaks
- Eliminate uncontrolled “shadow AI” tools
- Improve auditing and compliance
- Keep a clear record of who or what accessed sensitive systems
The company states that the platform can manage not only Microsoft-created agents but also those from other software providers, making it an AI governance hub for the entire organization.
Addressing the ‘Shadow AI’ Crisis
A major reason behind Agent 365 is the increasing problem of Shadow AI – employees using unapproved AI tools to speed up their work. While these tools can boost productivity, they also create significant risks due to unknown data handling, unverified security, and a lack of visibility for IT administrators.
By giving AI agents formal identities, governed access, and activity logs, Microsoft believes that businesses can finally bring these systems into compliance.
This approach resembles how companies dealt with the surge of mobile devices a decade ago when BYOD caused similar issues. Just like management solutions emerged to secure mobile workers, Agent 365 aims to protect the AI workforce.
A Massive New Market: 1.3 Billion Agents by 2028
Microsoft is not the only one recognizing the potential. The tech industry is quickly shifting toward AI agents that can take action, not just generate content.
Market analysts predict the number of active AI agents will reach 1.3 billion by 2028. This growth explains the excitement around agentic platforms. The market is expected to grow from $7.8 billion in 2025 to over $50 billion by 2030.
Big players are already positioning themselves:
- Google with Gemini Enterprise
- Amazon with Bedrock AgentCore
- Salesforce with Agentforce 360
Microsoft has a unique advantage: years of deep integration in corporate identity systems, device security, and productivity software. By embedding Agent 365 into the platforms companies already rely on, Microsoft makes it easier to adopt agentic structures.
Introducing Microsoft’s New Autonomous Agents
Along with the Agent 365 platform, Microsoft announced several new AI agents that demonstrate how these tools can integrate into business workflows.
Fully Autonomous Sales Development Agent
One of the standout announcements is a Sales Development Agent that can research prospects, qualify leads, and start outreach independently. Microsoft calls it a “new SDR team member” that requires no onboarding, training, or breaks.
The agent handles tasks traditionally done by entry-level sales staff, raising new questions and opportunities around the future of sales teams.
Security Copilot Agents
In cybersecurity, Microsoft’s Security Copilot gains new abilities that allow agents to:
- Draft or deploy Intune security policies
- Investigate alerts
- Summarize threat intelligence
- Run complex remediation workflows
For overwhelmed IT and security teams, Microsoft argues that these agents will act as tireless helpers, automating routine tasks.
Copilot Agents in Word, Excel, and PowerPoint
Microsoft is also enhancing agent capabilities in its productivity suite. Users can now ask Copilot to generate full reports, spreadsheets, or PowerPoint decks from scratch with simple chat prompts.
These agents can:
-
- Build slide decks based on topic descriptions
- Create long-form documents
- Compile structured data and analysis in Excel
- Maintain consistent style, structure, and formatting
This shift moves Copilot from an “assistant” role to that of an “autonomous creator.”
Windows ‘Agent Workspace’
To support agents running directly on devices, Windows is introducing a dedicated Agent Workspace — a secure, isolated space where AI agents can operate using their own IDs while being monitored by IT.
This workspace separates human and agent activities, preventing conflicts and allowing for clearer auditing.
Leveraging Microsoft’s Deep Roots in Enterprise Identity
Agent 365 builds on Microsoft’s vast experience in enterprise identity management. The company’s approach does not create a new system; it reuses the architecture that businesses already trust for authenticating millions of employees.
With agent identities in Microsoft Entra, organizations can apply the same Zero Trust principles to both humans and AI:
-
-
- Conditional access policies
- Privilege restrictions
- Compliance constraints
- Identity lifecycle management
- Behavioral analytics
-
This alignment between human and AI identities is a key part of Microsoft’s message: AI agents aren’t add-ons; they are new workers in the corporate team.
Rollout and Availability
Microsoft states that Agent 365 will be available starting this week through Frontier, its early-access program for new AI technologies. Pricing details have not been released, but industry analysts expect a subscription-based licensing model similar to current Microsoft 365 products.
Given Microsoft’s strong presence in the enterprise market, early adoption is expected to be high, especially among large companies dealing with unmanaged AI usage.
A New Era of Digital Labor
Microsoft’s Ignite announcements signal the start of a major change in business technology. The company is building infrastructure for a workforce of autonomous digital entities working alongside humans, rather than just tools to help humans.
If Nadella is right about a future measured “per agent,” then Agent 365 could become as vital to future businesses as Active Directory was during the age of networked employees.
AI agents are no longer just features. They are becoming workers, and now they have IDs to prove it.
