What happens when a critical CDN incident is escalating, and the answer is buried somewhere in billions of log rows? This discussion will dive into how a CDN veteran built an AI-powered observability practice from scratch, and why the hardest part might not be what you’d expect. Learn how to think about training an AI agent to understand your infrastructure, what it takes to get meaningful answers from natural language queries against massive log datasets, and how server-side CDN data can unlock insights that client-side monitoring alone will never surface. The speakers will also explore how those same techniques apply specifically to video streaming delivery — from manifest files to edge caching behavior — and what it might look like to optimize a streaming session in real time using AI at runtime.