Sylvain Kalache, Rootly
AI-assisted coding is exploding. Cursor is the fastest-growing SaaS company, and over a third of the Fortune 500 are using Copilot. But this acceleration doesn't translate into reliability. Quite the opposite.
This talk dives into the quirks of LLM-assisted development, with real-world examples: hard-to-trace bugs, AI-generated tests that mirror flawed logic, and hallucinated dependencies or slopsquatting that open up security gaps.
We'll examine the operational fallout: skyrocketing code churn, higher incident rates from shipping more changes faster, and large batch deployments that make debugging harder.
As developers become less familiar with their own code, leaner SRE teams are left to deal with the consequences.
We'll also explore the cultural shift: blaming outages on "the AI," eroding accountability. This talk offers actionable strategies for SREs, from adopting AI-powered incident tools to developing "incident vibing".
Sylvain leads AI Labs at Rootly, an initiative that aims to augment reliability engineering in the era of AI. Under his leadership, the lab has developed open-source prototypes, tools, and research collaborations, sponsored by organizations like Anthropic, Google DeepMind, and Google Cloud.
author = {Sylvain Kalache},
title = {From Vibes to Outages: Riding the {AI} Code Wave},
year = {2025},
address = {Dublin},
publisher = {USENIX Association},
month = oct
}