Most engineering orgs have adopted AI tools, but adoption rates alone don't translate to productivity gains. Data from LinearB's 2026 benchmarks shows high AI usage correlates with a 1.7–2.2x lift in PR merge rate, while light usage shows no measurable lift. Three engineering leaders — Angie Jones (Agentic AI Foundation), Balaji Srinivasan (LinkedIn), and Rebecca Murphey (Honeycomb) — explain that the real differentiator is leverage: how deeply engineers integrate AI into their workflows, not just whether they use it. Key insights include: the difference between enthusiastic adopters (task-by-task usage) and truly leveraged engineers (workflow-level improvements that benefit teammates), the bottleneck of agentic PR review capacity and how AI code review can address it, and why replicating top performers requires team-specific enablement on real work rather than broad exposure. All three leaders advocate replacing activity metrics like adoption rates, PR counts, and token usage with measures of delivered value.
