AI and Incidents with Sylvain Kalache

Gilles Chehade, co-founder and CTO of PlakarKorp, joins Matty to talk about why backups need rethinking now that ransomware is routine, AI agents run inside the network, and terabytes are the low end. They get into treating resilience like infrastructure as code, with an inventory, policies, and a reconciliation loop, and why restores are the part nobody tests.

Sylvain Kalache, who runs the AI lab at Rootly, walks Matty through what AI SREs can and can't do during an incident. They get into why handing every small incident to a robot erodes the practice responders need for the big ones, and what to measure so the humans on call don't burn out.

Jacquie Capur, a senior developer advocate at Nebius, joins Matty to talk about keeping meaningful control over the tools you depend on: knowing what they're tied to, verifying they do the job, and having a realistic way out. The conversation runs from configuration management's first magic moment to data residency rules in healthcare, tools that change every two weeks, and the case for open models.

Marino Wijay joins Matty to ask the question everyone in tech is quietly panicking about: what the hell is actually going on right now. They dig into the pace of change since the zero interest rate era, the collapse of infrastructure roles into a human-plus-AI layer, and why most of what looks like a bubble is really just an accounting hack that goes back decades.

Naga Sujitha Vummaneni and Sundeep Bobba join Matty to talk about their book, which reframes CI/CD pipelines as control systems: actuators, feedback signals, and constraints. They dig into what that model demands once AI agents start shipping code at machine speed, and why a bypassed control is a design signal, not a compliance failure.

Doug Pagnutti spent fifteen years as an automation engineer caught between the IT and OT worlds inside manufacturing plants. The friction sounds a lot like what's been happening in DevOps for over a decade. Matty and Doug dig into why industrial operations never got the memo, and what OT and IT could still learn from each other now that the wall between them is collapsing.

The software industry is experiencing another seismic shift. After decades of agile transformations, cloud migrations, and DevOps revolutions, we're now navigating the integration of AI into the software development lifecycle. But this time feels different—the pace is faster, the technology less mature, and the stakes feel higher. Matty Stratton sat down with Hannah Foxwell and Robert Werner to explore what this transformation means for developers, operations teams, and the organizations they serve.

Kat Cosgrove returns to talk with Matty about why security stays in the news, why most CVEs never get fixed, and which tools belong in every production environment. The conversation ends on burnout and the case for touching grass.

In this episode of Arrested DevOps, Matty and guest Kat Morgan discuss the ethical, practical, and technical implications of AI. They explore how AI can assist with coding, improve efficiency, and handle tasks, while emphasizing the importance of good practices and staying informed about the impact of AI.

Andrew Zigler of Mattermost and Matty talk about what makes an open-first developer community hard to run: the echo chamber, the pull of the paid staff, what can stay closed, and how to get more people advocating than just the ones with the title.