DORA 2025: AI Amplifies Team Dynamics Rather Than Fixing Broken Systems

Google Cloud's DevOps Research and Assessment (DORA) has released its State of AI-assisted Software Development 2025 report, warning that generative AI acts as an amplifier of existing team dynamics rather than a standalone productivity solution. The findings suggest that while AI accelerates high-performing teams, it often exacerbates dysfunction in struggling organizations.

Based on data from nearly 5,000 professionals and over 100 hours of qualitative research, the report highlights a stark divide in AI outcomes. While 90% of developers now use AI tools, only 17% have adopted autonomous agents, signaling a persistent trust gap in the technology. Most critically, the research found that for struggling teams, AI integration can lead to a 7.2% drop in delivery stability. The core conclusion is clear: AI does not fix broken teams; it amplifies what is already there.

The Shift Toward Systemic Productivity

DORA has historically focused on the "Four Keys" of software delivery performance. However, the 2025 research marks a strategic shift, moving beyond the evaluation of simple IDE plugins to examine the systemic organizational changes required for actual gains. To categorize these dynamics, DORA introduced an AI Capabilities Model and identified seven distinct team profiles. Among these is the "constrained by process" profile, which is specifically characterized by low individual effectiveness and high levels of burnout.

Prerequisites for AI Success

The report argues that the "magic wand" narrative of AI productivity is flawed. Instead, the data suggests that platform engineering and organizational health are essential prerequisites for AI success. DORA identifies seven critical organizational capabilities that determine whether AI will be a benefit or a burden: platform quality, data access, version control, the use of small batches, user focus, clear policies, and a defined AI stance.

Implications for Technical Leadership

These findings signal a necessary shift in strategy for CTOs and engineering leaders. Rather than investing in more sophisticated AI tools to solve performance issues, the report suggests that organizations must first prioritize fixing their delivery systems and improving data quality. Without these foundations, AI may simply accelerate the production of unstable code and increase team friction.

What to Watch

As organizations move from initial tool adoption to systemic implementation, the industry will be watching whether the gap between AI usage and the adoption of autonomous agents closes. The primary challenge remains whether companies can evolve their organizational capabilities fast enough to prevent AI from magnifying existing technical debt and operational instability.