AI coding agents write 30% more code but ship no more features
A study covering more than 700 software firms just put numbers to something a lot of developers have been feeling: AI coding agents make you write more code, but your team ships the same amount of software.
The Numbers
Introducing AI coding agents at a firm produces measurable output increases in raw code metrics:
- 30% increase in total lines of code generated
- 20% rise in total commits
- 23% more pull requests submitted on average
None of that translated into more features shipped. Resolution rates for Issues and Epics tracked in tools like Jira showed no statistically significant change after AI tools were introduced. The researchers also found no shift in the size or complexity of those tracked items.
Where the Time Goes
The culprit is code review. The average time between a pull request being submitted and it being merged into the codebase increased 49% after AI agents were introduced. The share of pull requests that required changes requested nearly doubled, and comments per pull request rose 35%.
The researchers describe this as efficiency gains in the coding phase being absorbed by downstream constraints. More code goes in; the review queue grows; the merged output stays flat. In response, firms increased the share of workers performing code reviews by 14%. The researchers found no significant employment reduction attributable to AI.
AI Review Is Not Filling the Gap
By March 2026, 80% of measured firms were using some form of AI code review. Despite that, AI agents accounted for only 23.3% of all review comments and 10.8% of all pull requests. Humans are still doing the heavy lifting on review, and the volume of work they are being asked to review has ballooned.
The Operator Takeaway
If you manage a dev team or hire contractors who use AI coding tools, the productivity gains you were promised are likely sitting in the review queue. The bottleneck is not generation speed. It’s human judgment on whether the code is actually correct. Until AI review quality improves enough to be trusted on complex pull requests, adding more AI coding capacity may just mean adding more review work.
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
