AI-Driven Ops Needs Reliability Before Autonomy

Datadog's Michael Whetten on Making AI Agents Production-Ready at Scale Rahul Neel Mani (@rneelmani) • December 8, 2025

Organizations are racing to deploy artificial intelligence agents in production environments to accelerate investigations and reduce manual effort. But the core challenge is reliability: AI's non-deterministic behavior often clashes with the consistency and accuracy required in high-stakes production environments.

See Also: Going Beyond the Copilot Pilot - A CISO's Perspective

Michael Whetten, senior vice president of product at Datadog, said success requires pairing AI's creative capabilities with rigorous validation. "It can do creative thinking, and it just has to be paired with logical thinking, such that if it comes up with generative hypothesis on something that could be wrong, how do you have the data and then know how to evaluate and validate or invalidate those hypotheses," he said.

In this video interview with Information Security Media Group at AWS ...