About the session
As organisations move from AI experimentation and copilots toward more agentic AI systems, governance challenges become more complex. Unlike traditional AI tools that mainly assist with content, analysis, or recommendations, agentic AI systems can plan tasks, call tools, trigger workflows, interact with enterprise systems, and influence operational decisions. This makes governance more important because risk can emerge during execution, not only during design or approval.
Traditional policy-based approaches remain necessary, but they are no longer sufficient on their own. In agentic environments, governance must increasingly operate at runtime, where AI systems are actually being used.
This session will explore how enterprises can translate governance and regulatory expectations — including accountability, human oversight, auditability, monitoring, traceability, and escalation — into practical runtime controls. The discussion will focus on control mechanisms such as access boundaries, data-use restrictions, tool-use controls, approval checkpoints, human review, logging, monitoring, and intervention mechanisms.
The session will offer a practical view of how enterprises can move from policy-driven governance to execution-level control frameworks that are better suited to agentic AI and growing regulatory expectations.
