81% of developers spend more time reviewing code after AI

The 2026 State of Engineering Excellence report from Harness found that widespread AI adoption is increasing software output while creating additional downstream work that many teams are not tracking.

Harness reports that 81% of developers say they spend more time on code reviews since their organizations adopted AI tools. About 28% of respondents said code reviews now take roughly 30% longer on average. The study estimates that roughly 31% of developer time is consumed by so-called “invisible work,” activities teams often do not record such as reviewing AI-generated code, fixing new bugs and switching between disparate tools.

The report identified a gap between technology leaders and frontline engineers on how well current metrics capture these changes. Some 89% of technology leaders continue to rely on productivity measures that do not reflect AI’s effects on individual developers or teams. Meanwhile, 94% of respondents said existing metrics omit factors such as technical debt and developer burnout. Managers were nearly four times more likely than frontline practitioners to report no concerns about how productivity is being measured.

When asked to name the single biggest challenge introduced by AI, respondents most often pointed to measuring true productivity impact (26%), maintaining code quality with AI (24%) and proving return on investment to leadership (18%). The report also notes an increase in manual remediation tasks created by AI outputs.

To address those gaps, Harness recommends that organizations revise how they track developer performance and treat AI output as a separate discipline. The report advises monitoring code validation and the quality of AI-generated code, tracking AI agent accuracy and acceptance rates, and accounting for tool costs separately from human output. Harness also reported that 49% of developers want to be involved in defining the metrics used to judge their work.

Trevor Stuart, SVP and general manager at Harness, said: “Cloud and the internet were infrastructure revolutions layered underneath the developer. AI is reshaping the developer’s job entirely, and the measurement frameworks that the industry has relied on for the past decade weren’t built for this new unit of work.”

The report summarized measurement as the biggest AI-related challenge facing engineering teams and recommended updated tracking and metrics to reflect new follow-up tasks and unrecorded labor.