What Good Looks Like: Quality Guardrails for Agentic Development

​Most of what makes a codebase good lives in developers' heads: the architectural decisions, the contracts between services, the sense of what "right" behavior looks like. That holds up until agents start writing most of the code, and no team can manually review thousands of changes a day. In this session, Vilhelm von Ehrenheim (Co-founder and CAIO at QA.tech) makes the case for codifying that knowledge into automated guardrails that run in your pipeline, spanning architecture, behaviour, and function. He covers why agents tend to erode maintainability when left unchecked, what's worth automating first, and how the teams moving fastest are already solidifying these systems today. A practical look at how to move fast and stay safe as agentic development scales.