In this episode of Software Leaders Uncensored, Steve Taplin sits down with Sylvain Kalache, Head of AI Labs at Rootly, to unpack what's really happening inside fast-scaling engineering organizations in the GenAI era. While most conversations focus on speed and productivity gains, Sylvain brings a much-needed perspective on what's breaking beneath the surface as companies adopt AI at scale.
Rootly, an incident management platform used by companies like NVIDIA, LinkedIn, and Figma, sits at the center of reliability engineering. That vantage point gives Sylvain a unique lens into how modern software systems behave once code hits production. And what he's seeing is both impressive and concerning.
AI Is Accelerating Development, But Speed Has Consequences
AI is undeniably accelerating development. Engineers are writing more code, faster than ever before, and teams are clearing backlogs at a pace that wasn't possible even a couple of years ago. At Rootly, this shift was so important that they actively tracked how much engineers were using AI tools. If someone wasn't leveraging them, leadership stepped in, not to enforce usage, but to understand why. The conclusion was clear. AI is no longer optional. It's foundational.
But speed comes with consequences.
Incident Rates Have Tripled Since 2023
Sylvain highlights a critical pattern emerging across the industry. As engineering velocity increases, so do system failures. More code changes mean more opportunities for things to break. In fact, he shared that incident rates across companies have tripled since 2023. Even organizations like Amazon are feeling the pressure, calling internal meetings to address rising outages tied to this new development paradigm.
The Bottleneck Has Shifted
This is where the real bottleneck has shifted. It's no longer writing code. It's everything that happens after. Code reviews, system reliability, and incident response are now under strain. Teams that once operated efficiently are struggling to keep up with the downstream impact of accelerated development.
AI for Reliability: The Rise of the AI SRE
Sylvain's role at Rootly's AI Labs is focused exactly on this problem. Rather than building large language models, his team applies AI to reliability engineering. One of their biggest innovations is an "AI SRE," an agent that can analyze logs, telemetry, and system data to identify root causes of incidents, sometimes before engineers even get involved. It's a direct response to the growing gap between development speed and operational capacity.
Engineers Are Becoming Managers of AI Agents
Another key insight from the conversation is how AI is reshaping leadership itself. Sylvain points out that engineers are no longer just individual contributors. They're becoming managers of AI agents. This requires a completely different skill set. Technical ability alone is no longer enough. Engineers now need to think in systems, coordinate intelligent tools, and make higher-level decisions about what should be built and how.
Context Replaces Constant Communication
At the same time, Rootly has taken a unique approach to maintaining alignment between teams. Because they've historically documented everything, from meetings to decisions, they can now use AI to compare context across product and engineering without needing additional meetings. Instead of pulling teams together to realign, AI agents analyze past conversations and ensure consistency. It's a glimpse into how organizations may operate in the future, where context replaces constant communication.
The Human Cost: Burnout Is the Real Risk
But perhaps the most important takeaway from Sylvain is not technical. It's human.
As AI tools become more powerful, the pressure on engineers is increasing. They're expected to do more, move faster, and adapt continuously. Sylvain warns that burnout is becoming a real risk, especially as incident volumes rise and the line between productivity and overload blurs. His advice to leaders is simple but often overlooked. Take care of your people.
The Companies That Win Won't Just Be the Fastest
The companies that win in this next phase won't just be the ones that adopt AI the fastest. They'll be the ones that understand its full impact. Not just on code, but on systems, operations, and the humans behind it all.
This episode is a reminder that while AI is transforming how we build software, it's also exposing new challenges that leaders can't afford to ignore.