[1/2] Atlassian’s State of Developer Experience 2025 (Category DevEx)

I studied an interesting Atlassian report on DevEx published last year. This is the second in a series launched in 2024 with the then-independent DX. In autumn 2025, Atlassian bought DX for $1 billion. The study used an online survey of 3500 developers and managers in six countries: the US, UK, Germany, France, Australia and India. It was conducted with independent firm Wakefield Research on 13–23 March 2025. Roles were split equally: 1 750 engineering managers or higher-level leaders and 1 750 engineers. Participants were recruited through email invitations.

Here are the main findings.

🤖 Widespread AI-tool adoption Almost all surveyed developers, 99%, now use AI tools and report time savings; 68% save more than 10 hours per week. In the previous 2024 report, most developers had not yet seen tangible benefits. The change over a year is dramatic.

🚀 How the saved time is used Developers put the hours saved into improving code quality, building features and writing documentation. They are trying to direct AI’s benefits toward product value through refactoring and functionality, and toward team support through documentation and knowledge sharing.

🤡 Substantial time losses persist despite the gains AI accelerates individual tasks, but organizational inefficiencies offset those gains. 50% of developers lose 10 or more hours a week to friction, while 90% lose at least 6 hours. On average, developers lose roughly as much time to obstacles as they gain from AI.

👾 The leading sources of friction Finding information about services, documentation and APIs; learning new technologies; and switching context between disconnected tools are the top 3 time drains. Working with other teams also makes the top five. Technical debt, previously the leading issue, has fallen outside the top 5 this year. Learning and frequently switching between new technologies stand out as emerging sources of friction.

👀 Coding is not the bottleneck Respondents spend only around 16% of their working time actually writing code, and coding itself is not among their main pain points. This explains why popular coding assistants can make life easier without fundamentally resolving productivity problems: the largest losses occur elsewhere. Accelerating coding has limited impact unless unproductive work is addressed alongside it.

💔 The gap between engineers and management The report finds a growing disconnect: 63% of developers believe leaders do not understand their daily challenges, compared with 44% a year earlier. The authors link this widening empathy gap to management’s eagerness to count AI time savings as a gain while leaving existing process obstacles intact. Leadership may overestimate AI’s impact while engineers remain overwhelmed by routine work and organizational problems.

In the next post, I’ll cover the authors’ main conclusions and recommendations.

#AI #ML #PlatformEngineering #Software #Architecture #Processes #DevEx #Devops