The Stack Overflow Podcast Your runbooks are obsolete in the age of agents
Oct 24, 2025
Spiros Xanthos, CEO of Resolve AI, dives into the transformative role of AI agents in incident management. He discusses how traditional runbooks are becoming outdated and how AI agents can autonomously troubleshoot complex systems, potentially saving up to 70% of engineers' time. Spiros elaborates on agents acting as on-call responders, gathering evidence across various tools, and assisting with the increasing complexity of software systems. He also touches on the evolving role of developers in an AI-driven landscape and offers insights into the future of engineering.
AI Snips
Chapters
Transcript
Episode notes
Pretrain And Integrate Common Tooling
- Pretrain agents on code, logs, metrics, and infrastructure so they can reason across data types out of the box.
- Connect common third-party tools first and add custom APIs or MCP descriptions for bespoke systems.
Agent Identified And Fixed Log Bloat
- Resolve identified high-volume log lines, traced them to code, and suggested specific code changes in minutes.
- A task that previously took hours was completed in two minutes with actionable fixes.
Closing The Loop From Detection To Remediation
- The next step is closing the loop: identify the problem, propose a fix, and let the agent enact changes when trusted.
- Trust and answer quality will determine how much remediation is automated versus human-approved.