87% say AI speeds up work, but only 28% see less friction

Atlassian published its State of Developer Experience Report 2026 on 20 October 2026, based on a survey of more than 1,500 software practitioners. It found that 87% of developers say AI helps them work faster, while only 28% say they experience noticeably less friction across their day-to-day workflow than before AI.

What the survey found

Atlassian reports that 80% of developers say AI helps them deliver more without working longer hours. The friction that has long slowed software delivery has stayed in place, though. Half of developers say they spend too much time finding information and attending meetings, and 45% point to context switching, according to ITPro's coverage of the report.

Checking AI output is a visible cost. Atlassian reports that 49% of developers spend four or more hours a week reviewing AI output, correcting prompts and similar work. Atlassian also reports that 75% of developers say they carry more responsibility for validating the quality and safety of AI-generated output.

ITPro notes that last year's version of the survey found 68% of developers saving at least ten hours a week with AI, and that the figure has fallen to 25% this year.

Strategy and measurement

Only 17% of developers and 23% of leaders say their organization has a clearly communicated, actively implemented AI strategy across the software development lifecycle, according to Atlassian. On measurement, 64% of leaders say greater AI usage correlates with better outcomes, and 41% of developers say they use AI more than necessary because they feel expected to. Atlassian's own summary is that AI usage is being treated as a productivity signal when it is not one.

Andrew Boyagi, Customer CTO at Atlassian, told ITPro that faster code is not the same as better software delivered sooner. Most AI use across the lifecycle still focuses on code generation rather than on finding information, clarifying requirements or moving work between teams.

What it means for engineering teams

The survey describes a shift in where engineering effort goes. Writing code gets faster, while reviewing, validating and coordinating take a larger share of the week. A team that measures AI adoption by usage alone can miss this, because usage says little about whether work reaches customers sooner.

Leaders who want a clearer picture can track outcomes such as lead time and review time next to adoption figures. They can also write down where AI is expected to help, since fewer than a quarter of leaders in the survey say their organization has a clear strategy.

What it means when you hire engineers

The 75% of developers who report more responsibility for validating AI output point to the skill that matters most in a hire: judgement about code they did not write. Engineers who use AI every day and still review its output with care add speed without adding rework for the rest of the team. Our how we vet process tests exactly this in a real codebase, and you can read more about our AI-native engineers.