Microsoft has recently rolled out three fundamental components tailored for AI agents. These components encompass Agent Connectors, which are underpinned by MCP (Model Context Protocol), along with Agent Workspace and Agent ID.

Agent Connectors empower developers to register application functionalities as agent connectors. This is achieved by constructing MCP servers and integrating them into the registry on the Windows device side. All this takes place within a securely isolated environment, ensuring data safety and system integrity.

Agent Workspace offers a policy-regulated, auditable, and isolated desktop environment. It enables agents to carry out tasks simultaneously without causing any disruptions to the user's primary session, thus enhancing multitasking efficiency.

Agent ID guarantees that every action performed by an agent is traceable and is kept distinct from user behavior, thereby maintaining clear accountability and enhancing security.

The relevant infrastructure is set to be accessible for public developer preview in the near future, providing developers with an early opportunity to explore and leverage these innovative features.