Topics

• Teaching AI to think like an SRE.
• Prototyping and iterating AI solutions.
• Human-agent collaboration.
• Evaluating AI effectiveness.

Key quotes / insights

• “AI’s biggest impact comes from compressing time — identifying causes minutes instead of hours.”
• Simple deterministic tagging + re-ranking often outperformed complex vector DB setups.
• “Time travel” evaluation — testing AI against known historical incident outcomes.
• Trust requires transparent reasoning and clear expression of uncertainty.
• Slack as primary human-AI collaboration interface.