AI-driven telemetry is up 10x. How to scale visibility to match it | Bill Emmett & James Conway
AI agents don't just consume telemetry, they generate it, and that is breaking both the economics and the architecture of how platform teams manage observability and security data. Bill Emmett, Sr Director, Technical Product Marketing at Cribl, and James Conway, Field CTO at Cribl, walk through what an AI-ready telemetry platform needs, from collection and in-flight normalization to federated search, data tiering and governance for agents. James also draws on Weave Intelligence's State of AI research, which finds that platform readiness, not model capability, is what's holding teams back from scaling AI: https://weaveintelligence.io/research/state-of-ai-in-platform-engineering-2026 0:00 Intro: how telemetry is changing in the AI era 2:22 Telemetry growth vs flat budgets 5:14 Solution-first vs platform-first tooling 8:39 Why AI agents break telemetry assumptions 10:21 4 requirements for AI-ready telemetry 15:18 The 6-capability checklist for agent-ready telemetry 20:49 Collect once, normalize and redact at the edge 23:59 Cutting noise and cost with dynamic routing 26:41 Federated search and querying data in place 28:41 Data tiering: caviar vs peanut butter 31:53 Cribl's AI platform for telemetry 40:37 Model routing to cut inference costs 42:13 3 levers for efficiency 44:19 Customer outcomes and avoiding lock-in 50:07 Q&A: Predictive telemetry 54:41 Closing thoughts for platform engineers Ready to build telemetry that keeps up with your agents? Keep going: 📘 Get your copy of the book, Thinking in Platforms: https://weaveintelligence.io/thinking-in-platforms-book 💬 Swap telemetry war stories with thousands of platform engineers in our Community Slack channel: https://platformengineering.org/slack-rd 🔔 Follow us on LinkedIn! https://www.linkedin.com/company/platform-engineering-org 🎓 Continue your platform engineering journey with Platform Engineering University: https://university.platformengineering.org #PlatformEngineering #Observability #AIAgents