What happens when your CTO builds a full-stack app almost entirely with AI coding agents?

CodiLime's CTO Krzysztof Wrobel did exactly that. In this session, he and senior engineer Paweł Gawliczek walk through the real story: what worked, what blew up in their faces, and the non-obvious lessons that took months of trial and error to figure out.

This isn't AI hype. It's an honest look at what using Claude Code in production actually looks like, from the first requirements doc to the final CI pipeline.

What you'll learn: Why AI makes engineering discipline more important, not less The "waterfall-ish" workflow they landed on after abandoning cowboy-agile AI coding Why keeping context small is the skill nobody talks about How to use quality metrics commit-by-commit to stop technical debt from quietly accumulating Their take on the hardest question in the room: what happens when a developer can't explain the code they shipped?

Plus a short live demo of Compass, the internal knowledge and competency management system Krzysztof built single-handedly using AI coding agents.

Whether you're already using AI coding tools or still figuring out if they're worth it, this session will give you a much more grounded perspective than most of what's out there.

CodiLime builds network and infrastructure software for some of the most demanding environments in tech.