State of AI Impact in Engineering:
Q3 Report
Download now
With AI-authored code at all-time highs, many engineering organizations are seeing faster code generation not translating into a proportional number of new features and capabilities.
Our Q3 data complicates the obvious story. Organizations generating high volumes of AI code wait 105% longer for a first review than those at the bottom, yet weekly PR throughput keeps climbing fast, up 75.5% over four quarters. Authoring capacity has scaled, and delivery volume has scaled with it. What hasn’t kept pace is everything that makes that volume safe to trust.
View insights
Throughput continues to grow
Growth of median weekly TrueThroughput, a complexity-weighted PR throughput metric, has held steady for the past four quarters, showing a 19.0% growth from Q2 to Q3 2026.
Median weekly TrueThroughput™
Sample from 500+ companies from October 2025 to September 2026
Review queues are saturated
Organizations with high AI code generation wait 105% longer for a first review than those with low levels of code generation. Review capacity hasn’t kept pace with authoring capacity.
Median time to first review by volume of AI code
Sample from 500+ companies from July 2026 to September 2026
Individual task relief outpaces systemic gains
When we look at developer sentiment towards different aspects of the daily software development workflow, we found a divide between where developers feel most and least supported. Developers are reporting high satisfaction with isolated, diagnostic tasks like production debugging, documentation and build and test, but marginal gains with systemic factors like code maintainability and clear project direction.
Percent change in DXI factors
Sample from 500+ companies from April 2026 to September 2026
About this data
DX’s Q3 2026 State of AI Impact in Engineering report publishes industry data from 500+ R&D companies. Data spans July-September 2026, except where noted.