AI Changed How Spotify Builds. What We Learned (and Fixed) About Quality at Higher Velocity

This post details how Spotify's AI-assisted development workflow has impacted quality and reliability. It highlights that while AI-authored code has not directly caused production incidents, the increased volume of change has outpaced some verification controls. The company is responding by strengthening safeguards, expanding rollback capacity, and scheduling automated changes during working hours. Compute shortages have necessitated a review of network edge and tiering approaches, and a strengthening of production service resilience. For mobile app quality, the focus is on adapting quality measurement systems to the faster pace of AI-assisted development, broadening quality signals, and incorporating longer-term trends. Data indicates that while total merged changes have more than doubled, quality and optimization work has also increased, and rework rate has not risen, suggesting no simple quality-for-velocity trade-off. However, code complexity and PR size are increasing, which are being monitored as potential warning signals. The post emphasizes that the primary challenge is ensuring the delivery system's controls operate at the same pace as development.