Agentic AI observability is now a first class production concern. Yanbing Li, Chief Product Officer at Datadog, joins Alan Shimel on Techstrong TV. Furthermore, she frames how AI agents force observability to grow up as a governance layer.

About Yanbing Li

Yanbing is a lifelong engineer who spent a decade at VMware. In addition, she led observability at Google Cloud and engineering for L4 self driving trucks at Aurora. Consequently, she brings a mission critical AI lens to product strategy. Furthermore, that background shapes how Datadog thinks about agentic AI observability today.

Inside agentic AI observability

Datadog frames agentic AI observability as more than a health check. As a result, the platform tracks the behavior of AI agents, not just uptime and latency. Meanwhile, agents act like distributed systems that touch data, APIs and models on every request.

Yanbing explains that observability now runs as one unified surface. Furthermore, that same surface powers autonomous detection, investigation and remediation. Consequently, the tool that watches AI is itself becoming an AI driven operator.

Why trust and data quality matter

Meanwhile, trust is the eminent question for production AI. Therefore, Datadog watches sensitive data, freshness, quality and lineage across the pipeline. In addition, a silent schema change upstream can make a healthy looking agent return wrong answers.

Yanbing also reports an inflection point at real customers. Furthermore, agent traffic on Datadog grew 30 times in the past 12 months. As a result, one enterprise cut incident response from hours down to about four minutes. Meanwhile, AI agents now handle roughly half of that customer's incidents on their own.

Explore more artificial intelligence coverage and the latest Techstrong TV interviews. In addition, Yanbing points engineers to hands on trials as the best way to feel how agentic AI observability works.

For more information please visit datadog.com