Fewer developers say AI saves them 10 hours a week

Last year, 68% of software developers said AI saved them ten or more hours a week. This year, only 26% say so. That comes from Atlassian's State of Developer Experience 2026, as reported by InfoQ. Over the same period, AI use became close to universal in engineering teams.

What changed

The report says 95% of developers and 96% of engineering leaders work at organizations that actively use AI in engineering. Most developers still see a local benefit. Eighty-seven percent say AI helps them ship faster, and 80% say it helps them deliver more without longer hours. But only 28% say daily friction is noticeably lower than before AI. Half say it is about the same.

Where the time goes

Atlassian's CTO for AI and Teamwork, Taroon Mandhana, said the bottleneck has moved to managing AI's work. The numbers fit that view.

• 49% of developers spend four or more hours a week managing AI tools.
• 65% spend time correcting AI work that missed the original intent.
• Among engineering leaders, those figures rise to 57% and 72%.
• 29% name reviewing or managing AI-generated code as a source of friction.

The older problems have barely moved. Finding information leads at 51%, then meetings at 50%. Coding has never appeared on the list.

The measuring problem

Sixty percent of developers say AI use is treated as a sign of productivity at their company. Forty-one percent say they sometimes use AI more than needed because they feel expected to. Only 42% of leaders say they have the right metrics to tell real gains from mere activity. One commenter suggested measuring accepted work, not usage.

Companies with a clear, actively used AI plan did better. Among their developers, 41% report less friction, against 22% where the plan is inconsistent or poorly explained. Only 17% of developers report such a plan.

Reasons for caution

This is research from a company that sells developer tools. It is based on self-reported views, not hard data, and Atlassian says the links it found are not proof of cause. The survey base also changed, from 3,500 people in 2025 to 1,560 this year. The report does not say whether the drop reflects more review work or simply higher expectations.