Sylvain Kalache · 2026-09-04 · notable
Sylvain Kalache — when AI handles incidents, engineers lose touch
Sylvain Kalache, AI Labs and developer relations lead at Rootly, argues AI incident response will pull average time-to-resolve down while making rare, complex outages take longer, because responders stop practising on the easy ones.
Kalache applies the 1983 'ironies of automation' argument to AI incident response: routine outages get easier, rare ones get harder.
What is it?
Sylvain Kalache's post argues that AI tools now clear routine incidents well enough that engineers stop building hands-on familiarity with their own systems. He frames it as a trade rather than a win: common failures resolve faster, while the rare novel outage that automation cannot handle takes longer because the humans are out of practice.
How does it work?
The argument rests on Lisanne Bainbridge's 1983 paper 'The Ironies of Automation', which found that automation removes an operator's chance to practise routine work while leaving them responsible for new and abnormal situations. Kalache carries the aviation parallel across: commercial pilots may see fewer than one engine shutdown per 100,000 flight hours, yet still sit recurrent simulator training every six months.
Why does it matter?
The recommendation is concrete — build regular incident simulations and hands-on practice into on-call readiness instead of reading a falling MTTR as proof the team is ready. Kalache predicts average time-to-resolve will drop for most incidents while resolution time rises for complex ones. He leads AI Labs and developer relations at Rootly, a company that sells incident tooling, and was previously a senior SRE at LinkedIn.
Who is it for?
SRE and on-call teams