India, Sept. 29 -- Enterprise AI is entering a more difficult phase. The first wave largely involved systems that assisted employees with recommendations, analysis, or content. Agentic AI changes that equation. Agents can increasingly act on enterprise systems, interact with infrastructure, generate machine data, and participate in security and IT operations.

Now that creates a different risk. Enterprises are giving AI more autonomy at a time when many are still working out what those agents can access, what actions they can take, how much they cost to operate, and when humans need to intervene.

For Cisco and Splunk, that is pushing observability beyond monitoring applications and infrastructure. The emerging requirement is to understan...