This is Fine! A podcast about resilience engineering and software SRECon Americas 2026 recap
15 snips
Apr 14, 2026 They recap standout conference talks on disaster recovery, resilience engineering methods, and alternatives to root cause analysis. They touch on AI hype, measurable productivity claims, and agentic tools for incident investigation. Stories about documentation, teaching teams to learn, and hands-on incident analysis exercises make appearances. Breathing-based recovery techniques and hiring practices for SRE roles are also discussed.
AI Snips
Chapters
Books
Transcript
Episode notes
Agentic Coding Produced A 600-Line Diff
- Colette described using agentic coding to generate a 600-line diff and then manually reviewing and reasoning about whether it did what she intended.
- She compared that review role to code reviewers: agents can produce code but humans must validate intent and correctness.
Always Pair Metrics With Their Limitations
- When using metrics like DORA or MTTR, always surface their limitations to consumers so leaders understand what the numbers actually mean.
- Clint warned executives love neat metrics, but teams must pair numbers with rich system stories to avoid misleading conclusions.
AI Productivity Claims Can Drive Misleading Investment Loops
- Both hosts observed an AI hype feedback loop: teams claim large productivity gains (40–60%) without rigorous measurement, which drives more investment.
- They cautioned this amplifies misinterpretation and could misdirect scarce GPU/cloud budgets.
