DORA’s 2025 research on the impact of AI

Nathan Harvey and Laura Tacho kick off the discussion by addressing the limitations of relying solely on the four key DORA metrics for measuring AI impact. They highlight that while these metrics provide a basic perspective on software delivery performance, they are only leading indicators and insufficient for deeper insights. Metrics alone cannot account for the complexities of software delivery, and there are additional factors in play that leaders need to consider for a comprehensive view.

The conversation progresses to the evolution of DORA metrics, noting that for the past two years, research has expanded to include five software delivery performance metrics instead of just four. Leaders are advised to look beyond just these metrics and consider organizational health indicators, which will help them assess the real impact of their AI strategy.

AI is described as an amplifier that can highlight both strengths and weaknesses within engineering systems. If a team has efficient delivery processes, AI can significantly enhance productivity. Conversely, if dysfunction exists, AI will amplify these bottlenecks and issues instead of resolving them. Understanding how this amplification works is crucial for leaders managing AI integrations.

The discussion touches on the differing patterns of AI usage between junior and senior engineers. A significant finding is that junior engineers tend to engage more actively with AI tools, leading to more code generation. However, this trend places greater pressure on the code review process, necessitating an analysis of team dynamics and the operational pressures that come with increased AI-induced workloads.

Nathan shares insights on the DORA AI capabilities model, which includes characteristics essential for maximizing AI benefits in software development. These characteristics, such as having a clear AI stance, healthy data ecosystems, and strong version control, help teams utilize AI effectively while minimizing friction. The model emphasizes that successful AI integration is contingent upon how organizations use these technologies.

Towards the end, both Nathan and Laura stress the importance of open communication channels within teams. Direct conversations about friction points and challenges faced by teams can provide leaders with invaluable insights into operational bottlenecks, ensuring that AI is utilized in ways that reduce friction and enhance productivity.