loading
loading
Atlassian’s 2026 State of Developer Experience report reveals a disconnect between AI adoption, developer productivity and organisational performance. An engineering leader recently asked me how AI could help their team work more effectively, to keep up with a rapidly growing workload. It turned out the problem was upstream teams creating significantly more output with the help of AI. Using AI to battle the influx of AI generated work would pass the bottleneck down to the next team without solving the problem. That conversation came to mind while reading Atlassian’s 2026 State of Developer Experience report. 87% of developers say AI helps them ship work faster, but 72% experience the same or more friction than before AI. Improving Developer Experience, and as a byproduct, productivity, has always been about removing friction from developers daily work. We’re presented with an interesting scenario to examine – simultaneous increases to both developer output and friction. Accelerating one part of a bigger system doesn’t help Working with the Atlassian Williams F1 Team is a constant reminder to think in systems. Improving one part of a Formula 1 car doesn’t necessarily result in a faster car; it can create new constraints elsewhere resulting in worse performance. The goal is a better performing car, not a better performing part. It’s the same in software development, optimising for one part of software delivery, coding, creates new problems downstream as more work needs to squeeze through the same bottlenecks. The goal is higher quality software delivered faster, not more code faster. This was the case for the previously mentioned team, AI caused a huge increase in workload but the downstream constraints prevented that being converted into faster value. Even with faster output, time savings are falling: in 2025 68% of developers said they saved 10+ hours a week, compared with 25% in 2026. This is surprising given the rapid improvement in models. It’s worth questioning how much of the time gained through faster coding is now consumed elsewhere. The report shows 65% of developers now regularly spend time correcting AI output. Leaders should be careful about optimising work in one place while creating more work in other places. Making one task faster doesn’t improve the whole system. Coding is fast, what’s slowing things down? The same challenges have been frustrating developers for years, and despite heavy investment in AI there isn’t much movement in the top friction points for developers. Finding information takes top spot again in 2026 as the biggest source of friction, followed by meetings, context switching, technical debt and unclear requirements. It raises a question I’ve been asking technology leaders for years: How much capacity goes into coordinating work, compared with doing the work? Developers spend time searching for information, understanding priorities, finding the right people, navigating dependencies and attending meetings to stay aligned. Organisations are getting better at managing this, but managing friction isn’t the same as removing it. Improving Developer Experience has always been about understanding friction points and systematically addressing them. We saved a group of 150 employees at Atlassian Williams F1 over 863 hours in low value meetings per month without using AI. We did this by deeply understanding the root cause of meetings and establishing practices to address them. This is where AI has enormous potential. AI can help developers find information, understand dependencies, reduce repetitive tasks and move work across the organisation more effectively. These are closely connected to the biggest sources of friction identified in the report and addressing them could unlock significantly more value than faster coding. AI adoption is not an AI strategy The most concerning finding in this year’s report relates to how organisations are approaching AI. Only 17% of developers and 23% of leaders say their organisation has a clearly communicated and implemented AI strategy. 60% of developers say AI usage is treated as a productivity signal, and 41% sometimes use it more than necessary because they feel expected to. Let that sink in. Many organisations haven’t clearly defined what they want to achieve with AI, but they’re measuring how much people are using it as a proxy for productivity. Developers know they’re being measured and feel obliged to use AI even in scenarios in scenarios where it might not add value. Not only is this an expensive mistake, consider the impact on the broader system. Think back to the team struggling with increased work flowing into their backlog. The upstream teams might be considered “productive”, even though their potentially unnecessary usage of AI causes the downstream teams to struggle, and slow down the rest of the system. It’s a completely different story for developers at organisations that have a clear AI strategy. They’re twice as likely to experience less friction than before AI was introduced. This suggests that organisations that are deliberate about the problems they want AI to solve are more likely to realise its potential. The real opportunity I’m incredibly optimistic about the impact AI can have on developer experience, but its greatest potential goes far beyond speeding up individual tasks. To unlock this value organisations will need to rethink how work gets done, remove friction between teams, and free teams up to work on higher-value work. Realising the opportunity requires leaders to be deliberate about the problems AI can solve, and measure the resulting outcomes, not activity. The goal isn’t to generate more work, it’s to get more of the right work done with less friction. Join me on November 9 I’m going to discuss these results next month on November 9, with Kelsey Hightower (Author, Investor, and Open Source Contributor), Nathen Harvey (Principal Technologist, AWS), and Jennifer Riggins (Technology Journalist). We’ll dig into why so few organisations have an AI strategy their teams can see, how to measure without rewarding performative AI use, and how leaders can continue to upskill their teams, even as roles evolve with AI. Bring your questions. I’d love to hear what you’re seeing in your own teams.
document fetched 2026-10-24 11:26 · canonical https://atlassian.com/blog/development/developer-experience-report-2026