Microsoft Agent 365 is the software giant’s new platform for deployment, management and control of AI agents. It has been presented at Ignite 2025 and confirms the total commitment to agency AI.
Microsoft is holding its professional conference this week and, as planned, new artificial intelligence technologies occupy everything. Or almost everything, because the company has also made other announcements of interest such as the expansion of the Cloud PC cloud computing service or the new tools for PC recovery, which are part of the ‘Resilience Initiative’ that Microsoft launched after the very serious CrowdStrike case of 2024 that left millions of Windows machines unable to start.
What is Microsoft Agent 365
It is known that AI agents (or agentic AI) are one of the main technological trends of 2025. A revolution within the revolution of the era of artificial intelligence that invades us, with the potential to generate impressive income and at the same time obstacles to its effective implementation which are also of enormous dimensions.
In fact, reports from large consulting firms on agentic AI speak of ‘pure hype’ and assure that most AI projects with agents remained early-stage experiments or proof-of-concept demonstrations, driven more by publicity than by practical results.
Microsoft Agent 365 wants to break that perception, accelerating the creation of a «agent factory» of AI that allows companies to manage their own agents. While Microsoft was founded with the idea of being a software factory, it is rapidly transforming into an era where AI agents will take on more and more human tasks.
Precisely, the new platform aims to allow companies to manage AI agents in the same way they manage their human workers, says Jared Spataro, marketing director of applied AI: «Agents are already transforming the way we work and IDC (in a report sponsored by Microsoft) predicts there will be 1.3 billion agents by 2028… Agent 365 is the agent management platform, extending to agents the infrastructure you trust to manage your staff».
AI assistants: equal to or more than humans
It must be said that Agent 365 is not a platform for creating enterprise AI tools; It is a way of managing them, as if they were human employees. And the biggest current challenge, according to experts, is not to create AI agents, but to organize and govern those that already exist.
Companies that use generative AI agents in their digital work environment can use Agent 365 to organize their growing botnet, monitor their performance, and adjust their configuration. In essence, Microsoft has created a trackable workspace for agents: «The tools that are used today to manage people, devices and applications will want to be extended in the future so that they can also be used to manage agents»says Charles Lamanna, president of Copilot, Microsoft’s AI chatbot.
Lamanna foresees a future in which companies have many more agents than people performing tasks. For example, if a company has 100,000 employees, it estimates that it will use between half a million and a million agentswho will perform everything from simple tasks such as organizing email to comprehensive management of the purchasing process. The executive assures that Microsoft uses millions of agents internally.
This army of bots, with permission to perform actions within the company’s software and automate aspects of employee workflow, could be difficult to manage. The lack of clear oversight could also expose companies to security breaches. Agent 365 allows you to manage all bots, regardless of whether they were created with Microsoft tools or through a third-party platform.
Microsoft Agent 365 features and capabilities
The primary function of Agent 365 is to serve as a centralized registry of an organization’s active agents, with specific identification numbers for each and details about their use by employees. It also allows you to modify agent settings and access permissions to different parts of the business software.
The tool includes security measures to analyze the activity of each agent in real time. As data flows between people, agents and applications, security will have to be increased. As more companies implement pilot programs to test AI agents, more questions are being raised about the security of this technology when integrating it into essential workflows that often contain sensitive data.
Agent 365 works with agents built with Microsoft tools (such as Copilot Studio or Microsoft Foundry) as well as third-party tools built with open source frameworks.
Microsoft highlights five key capabilities from Agent 365 as a summary:
- A registry that provides you with a single source of truth for all agents in your organization.
- Access control to manage agents and limit their access to only the resources necessary to perform specific tasks.
- Visualization: A unified dashboard and advanced analytics to see connections between agents, people, and data, and monitor agent behavior and performance in real time.
- Interoperability to provide agents with applications and data that simplify workflows between humans and agents.
- Security to help protect actors from threats and vulnerabilities, and to detect,
investigate and remedy attacks directed against them.
Microsoft Agent 365 is now available for testing in the early access program created by the company to introduce the new AI functions and called ‘Frontier’.
