The DX Core 4 framework (speed, effectiveness, quality, business impact) remains a stable measurement architecture for engineering productivity even as AI coding tools reshape how software is built. Rather than inventing new AI-specific metrics, engineering leaders should treat AI telemetry (adoption rates, token usage, agent task counts) as diagnostic context that explains how work is changing, not as a replacement for outcome-oriented measurement. Traditional metrics like PR throughput still signal engineering system flow, but their interpretation must evolve because AI changes the behaviors that generate them. For example, a high PR merge rate in an agentic workflow could mean excellent code quality or rubber-stamping of AI output. The recommendation is to triangulate across diagnostic, system, and outcome metrics to understand whether AI investments are actually delivering results.
