AI NewsProductivityAnnouncement

Atlassian survey says AI time savings fell from 68% to 25%

Atlassian's 2026 State of Developer Experience report says time savings from AI fell from 68% to 25% in a year. 65% of developers now regularly correct AI output.

AI News

Editorial3 min read

LinkedInX
Cover image for Atlassian's 2026 State of Developer Experience report

Image: Atlassian

Why it mattersA team using AI code tools no longer gets ten free hours a week. Review and correction take most of the gain. Measuring AI use as a productivity signal rewards activity over finished work.

Many engineering leaders still plan for AI to give each developer a day back each week. Atlassian's new survey says that gain is mostly gone. In 2025, 68% of developers said they saved ten or more hours a week with AI. In 2026, 25% say so.

That figure comes from the 2026 State of Developer Experience report. Atlassian's own blog post, by Head of DevOps Evangelism Andrew Boyagi, carries the figure. The report surveyed 1,155 individual contributors and 405 engineering leaders. InfoQ also covered the release.

What the report says is still working

AI adoption is now near universal. The report says 95% of developers and 96% of engineering leaders say their organisation uses AI in engineering. Local wins have not gone away. 87% of developers say AI helps them deliver work faster. 80% say it helps them deliver more without longer hours.

What has not changed is the top sources of friction. Finding information leads at 51%, followed by meetings at 50%, context switching at 45% and technical debt at 44%. Finding information topped the 2025 list and placed second in 2024. Coding has never appeared on the list.

Where the time goes now

Correction work takes a large share. The report says 65% of developers now regularly spend time fixing AI output. 49% spend four or more hours a week managing AI tools. Both numbers run higher among engineering leaders.

Taroon Mandhana, Atlassian's CTO for AI and Teamwork, summarised the shift in one line: the bottleneck has moved to managing AI's work. Boyagi's post names the real goal as higher quality software delivered faster. Code volume is a side effect of that goal rather than a target.

Measurement rewards the wrong thing

The sharpest finding is about how companies measure AI. 60% of developers say AI usage is treated as a productivity signal at their organisation. 41% say they sometimes use AI more than necessary because they feel expected to. Only 42% of leaders say they have the right metrics to tell real gains from activity.

Clear strategy is still rare. Only 17% of developers and 23% of leaders say their company has a clearly communicated and actively used AI strategy. In those companies, 41% of developers report less friction than before AI, against 22% at companies without one.

Caveats sit on the headline. This is vendor research from a company selling developer tools. InfoQ adds that the sample base changed too. The 2025 edition surveyed 3,500 developers and managers. This year surveyed 1,560. Neither edition publishes fieldwork dates. Atlassian itself writes that its findings reflect self-reported experience, with no system-level telemetry behind them. The company also writes that associations in the data should not be read as causal proof.

The number worth carrying out of this is the 65% who correct AI output every week. Review capacity sets the limit now that coding is faster. A team measuring AI use by tokens or by lines accepted misses that limit. Measuring accepted work, finished features and defects caught in review tracks the system Atlassian describes.

Source

SourceAtlassian

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.

Share
LinkedInX
Start a project