BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.devopsdays.org//dodroxrox26//talk//HKGHAN
BEGIN:VTIMEZONE
TZID:US/Mountain
BEGIN:DAYLIGHT
DTSTART:20250922T000000
TZNAME:MDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0600
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251102T020000
RDATE:20261101T020000
TZNAME:MST
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20260308T030000
RDATE:20270314T030000
TZNAME:MDT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:AI Made Developers 25% More Productive. It Also Tripled Our Incide
 nt Rate. - Sylvain Kalache
DTSTART;TZID=US/Mountain:20260922T140000
DTEND;TZID=US/Mountain:20260922T143000
DTSTAMP:20260729T024951Z
UID:pretalx-dodroxrox26-HKGHAN@talks.devopsdays.org
DESCRIPTION:Developers using GenAI tools ship 25% more tasks\, accept a bi
 llion lines of AI-generated code per day\, and are more productive than ev
 er. So why are production incidents up 3x?\n\nThis session shares what we 
 found after analyzing incident trends across thousands of engineering team
 s and why the productivity gains from AI-assisted coding are creating an i
 nvisible reliability tax that most organizations aren't yet measuring.\n\n
 We'll walk through the Incident Rate Formula (λ + C × P): as AI accelera
 tes the rate of change (C)\, the probability that any given change causes 
 failure (P) becomes the only lever infra engineering teams can control. We
 'll look at how AI actually fails in production\, echo chambers in unit te
 sts\, slop squatting attacks\, hallucinated dependencies\, and why the "an
 gry intern" mental model is the most useful frame for working safely with 
 LLMs.\n\nThe talk closes with practical strategies for keeping P low: muta
 tion testing at scale (Meta's 70%+ acceptance rate)\, AI-augmented code re
 views\, MCP-powered incident investigation\, and where AI SREs are already
  handling Sev-2 and Sev-3 incidents autonomously.\n\nIf you're building wi
 th AI or leading teams that are\, this talk will give you a data-driven fr
 amework for staying fast without letting reliability quietly collapse.
LOCATION:Main Stage
URL:https://talks.devopsdays.org/dodroxrox26/talk/HKGHAN/
END:VEVENT
END:VCALENDAR
