Microsoft CEO Satya Nadella has proposed that AI agents should be managed like human employees, with distinct identities and permissions. Speaking on the 'Possible Podcast,' he highlighted the challenges companies face as they integrate AI into their operations. Nadella explained that Microsoft is exploring necessary tools and policies to oversee its numerous AI agents effectively. This includes establishing specific permissions and auditing capabilities for these agents. To facilitate this, Microsoft has launched Agent 365, which encompasses tools like Entra for digital identity management and Purview for data labeling. Nadella also noted the cognitive burden he experiences while managing up to 100 AI coding agents simultaneously, emphasizing the complexity of this task. He concluded that ensuring security, containment, and observability will be vital for fostering confidence in the use of AI agents within organizations.