Incident Response and DevOps in the Age of Generative AI
Heavybit by Jesse Robbins · · Article
"GenAI is good at confidently delivering text that is pleasant to read, but not always complete, or correct."
Jesse Robbins convenes a panel of incident management veterans to examine where generative AI genuinely helps in SRE and DevOps — and where humans must stay in the loop.
Jesse Robbins convenes a panel of site reliability veterans to examine generative AI’s genuine role in incident management and DevOps — where it helps, where it fails, and why humans may always need to stay in the loop. The panel draws on more than 40 years of collective experience from practitioners at Jeli, PagerDuty, AWS, and other leading infrastructure companies.
The article explores GenAI’s core tension in operational contexts: the technology is capable of generating confident, readable output — but “not always complete, or correct,” and not equipped to make consequential decisions under uncertainty. Experts including Nora Jones, Jeremy Edberg, Mandi Walls, and Brent Chapman discuss where AI-assisted tooling reduces toil in incident workflows versus where ad hoc collaboration and seasoned human judgment remain irreplaceable. Robbins frames the question as one of division of labor: deferring low-level tasks to AI so engineers can focus on strategy, not replacing the humans who understand context and consequences.
Also Mentioned
