Speed Is Easy Trust Is Hard (and Other Surprises from AI at Scale)

AI adoption is accelerating across engineering organizations, but translating that adoption into meaningful impact remains uneven. While many teams see immediate gains in speed, those gains often stall as new bottlenecks emerge around verification, coordination, and trust. This talk shares lessons learned from observing AI adoption at scale across large engineering systems. We will explore what actually drives effective usage, why increased output does not always translate into better outcomes, and how organizations can rethink workflows to fully realize the benefits of AI.

We will cover practical strategies for driving adoption beyond surface-level usage, approaches for handling verification in AI-assisted development, and ways to measure success that go beyond activity metrics. The goal is to provide a clear, experience-backed framework for turning AI from a promising tool into a reliable driver of engineering impact.