Developers Reporting Double-Digit Weekly Time Savings from AI Fell from 68% to 26%

A significant shift has emerged in developer workflows, as Atlassian’s State of Developer Experience 2026 reveals that the percentage of developers saving ten or more hours weekly using AI decreased from 68% to 26% year over year.

While adoption nears saturation—with 95% of developers and 96% of engineering leaders reporting organizational AI use—engineering leaders note they are now dedicating more time to refining and correcting AI outputs than developers are.

Despite this, the practical benefits haven't vanished entirely; eighty-seven percent of developers feel AI helps them ship faster, and 80% report delivering more without extending their workday.

However, relief from daily hurdles has diminished, with only 28% reporting noticeably less friction than before AI, while 50% describe a friction that is "of a different type than before AI."

This new friction stems partly from the AI tools themselves: 49% of developers now spend four or more hours weekly managing AI tools, and 65% report correcting AI-generated work that missed the original intent.

Among engineering leaders, these burdens are heavier, with 57% spending time managing AI and 72% correcting AI output. Furthermore, 29% cite reviewing or managing AI-generated code as an outright source of friction.

The report highlights that 60% of developers feel AI usage is treated as a productivity signal within their organizations, leading Shan Huang to suggest that if usage is driven by expectation, the issue is a "metric problem," not a tooling one.

Taylor Johnson offered a concise alternative perspective: "AI activity is not impact; measure accepted work, not usage." Organizations possessing a clear, implemented AI strategy show better results, with 41% of their developers reporting less friction compared to 22% in those with inconsistent strategies.