archive – Shane Murray
summary
Claims that moving gen AI past pilot reveals a persistent illusion of done and day‑two failures around reliability, observability, and trust. Cites survey numbers: 100% feel pressured, 91% deploy or are deploying, 68% say data is not AI ready. Describes a troubleshooting agent that reduces time to resolve incidents by ~80%, is built in AWS Bedrock with Claude 3.5 Sonnet, and returns recommended root causes and steps in under a minute.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:16Generative AI illusion of done
- 3:42Generative AI fundamentals
- 6:33Monte Carlo troubleshooting agent
- 9:27AI applications lessons
- 11:59AI production reliability pains
- 15:11Operationalizing AI reliability
- 17:42Summary
related talks
50mAI IN SRE: HYPE VS REALITY IN 2026
19mAI Agents for Incidents: The Good, the Bad, and the Ugly
29mAI Conversation
01h 07mNew Methods of Working with AI.
32mScaling a Responsible Agentic Workforce
30mArchitecting Integrity: Responsible Software Evolution in the AI Era
29mHow to use AI to fight your DevOps transformation!
29mThe AI Teammate Is Here — Now What?
