A Plandek analysis of 2,000 organizations finds that lower-performing software engineering teams gain nearly 50% improvement in lead time to delivery after adopting AI coding tools, a rate four times greater than high-performing teams. Despite this, bottom-quartile teams still lag significantly: they take over 35 hours to merge pull requests vs. under 21 for top performers, ship in 62+ days vs. 22.5, and complete less than half their planned sprint work. The gap persists because AI helps lower performers adopt best practices more easily, while high performers face systemic workflow constraints AI alone cannot fix. Looking ahead, the report suggests future engineering teams may shrink to one or two engineers augmented by AI agents, with the overall pool of organizations able to build custom software expanding as AI lowers the talent barrier.
Sort: