Microsoft CEO: Treat AI Agents Like Employees, Not Software
Satya Nadella says companies need formal management systems to oversee autonomous AI agents as they take on more complex workplace tasks. The Microsoft CEO described overseeing around 100 AI coding agents as mentally demanding, signaling that businesses can no longer treat these systems as conventional tools.
As AI agents become more capable of executing multi-step tasks with minimal human input, organisations face new governance challenges. These systems now schedule meetings, book travel, update records, process information and interact with digital services on behalf of users-work previously handled by people.
What Nadella Says Companies Need
Organisations should establish three core controls:
- Assign digital identities to each AI agent
- Restrict their operating environments and access permissions
- Create clear policies governing their behaviour and data access
These measures are essential as AI systems move from assisting users to independently executing decisions, Nadella said. Without them, companies risk unintended actions, inaccurate execution and unauthorised access to sensitive information.
The Business Opportunity and Risk
Companies increasingly view autonomous agents as productivity enhancers. They can handle routine work at scale, freeing staff for higher-value tasks. But that same autonomy creates accountability gaps-organisations need visibility into what agents do and why.
Nadella stressed that confidence in AI agents depends on governance mechanisms as strong as those used for critical business processes. Organisations must be able to observe, contain and manage agent behaviour in real time.
Microsoft's Response
Microsoft is developing tools that combine identity management with information governance to help companies track, secure and monitor AI-driven activities. The goal is to give organisations the oversight infrastructure they need as agent deployment accelerates.
For managers, this shift means adding AI agent supervision to existing oversight responsibilities. The framework Nadella describes treats autonomous agents as workforce members requiring the same governance attention as human employees-identity verification, access controls and accountability measures included.
Learn more about AI for Management and AI Agents & Automation to understand how these systems fit into organisational structures.
Your membership also unlocks:
