Rapport DORA 2025

2025-09-23: «Google Cloud» published the «Rapport DORA 2025», subtitled "State of AI-Assisted Software Development", with «Derek DeBellis» and «Nathen Harvey» among its authors. Its evidence base is close to 5,000 surveyed technology professionals plus more than 100 hours of qualitative research.

The finding that keeps resurfacing is a split one. AI adoption is near universal (90% of respondents use at least one platform, and more than 80% believe it raises their productivity), yet the report measures a positive correlation with delivery throughput and a negative one with stability. The authors are explicit that these are correlations, not causalities. That asymmetry is what makes the document quotable: it documents a cost alongside the speed, without claiming to have found the mechanism.

The report's central framing is that AI amplifies whatever team dynamics already exist. A team with weak development practices, absent documentation, or siloed culture sees those problems intensified; a team with solid technical fundamentals sees its capabilities multiplied. From that reading, DeBellis and Harvey identify seven team archetypes and seven capabilities that amplify AI's positive effect, from clarity of AI policy and connection to internal context through to platform engineering and outcome measurement. Their recommendation to leaders follows: treat AI adoption as an organizational transformation rather than a tool deployment.

SFEIR's 2026 note on SDLC and PDLC reuses the same throughput-versus-stability result as evidence that the bottleneck has moved upstream, toward deciding what to build.