News

AI has the New Baseline: What Google Cloud's 2025 DORA Report Means for Developers

• By John K. Waters
• September 24, 2025

In 2025, AI isn't just in the developer toolbox, it's the foundation of modern software engineering. Google Cloud's 2025 DevOps Research and Assessment (DORA) Report makes it plain: 90% of technology professionals now use AI at work, up 14% from 2024. Developers spend a median of two hours per day working with AI tools. Nearly two-thirds rely on AI for at least half their workflow, and one in twelve say their development work is now almost entirely AI-mediated.

The outcome? Eighty percent report improved productivity, and nearly 60% say code quality is up. But there's a catch.

The Trust Paradox for Developers
Despite the measurable gains, trust in AI remains shaky. Only 25% of respondents say they trust AI outputs "a lot" or "a great deal," while 30% admit to "a little" or "not at all."

For developers, this means AI is being treated less as an autonomous coder and more as a productivity accelerator. AI outputs are funneled into pipelines guarded by version control, test automation, and human oversight. The mantra is "trust but verify."

Throughput Gains, Instability Risks
DORA continues to track software delivery along two key axes: throughput (deployment frequency, lead time, recovery from failure) and instability (change fail rates, rework).

AI boosts throughput: teams ship faster and recover more quickly. But it also introduces instability, especially for organizations without strong delivery scaffolding. The report finds that high-performing teams are those investing in AI adoption and the cultural and technical practices that sustain reliability.

Seven Archetypes in the AI Era
To capture the landscape, DORA clustered teams into seven archetypes. At the top are "Harmonious High-Achievers" (20%), balancing throughput and stability with low burnout. At the bottom are teams with "Foundational Challenges" (10%), where process gaps and high burnout undermine AI's potential.

Other archetypes include "Stable and Methodical," "High Impact, Low Cadence," "Legacy Bottleneck," and "Constrained by Process." The taxonomy highlights that AI reflects organizational realities rather than reshaping them on its own.

The DORA AI Capabilities Model
New this year is the DORA AI Capabilities Model, which outlines seven organizational practices that separate AI success stories from the rest:

• Clear AI stance and governance policies
• Healthy, accessible data ecosystems
• AI-accessible internal data for context-rich coding help
• Strong version control practices
• Small-batch, iterative workflows
• User-centric product focus
• Quality internal developer platforms

For developers, these aren't abstract management goals. They shape whether AI makes life easier—or introduces more firefighting.

Platforms, Pipelines, and Value Streams
The report underscores the industry-wide embrace of platform engineering. Ninety percent of organizations now run at least one internal platform, and three-quarters have platform teams.

High-quality platforms reduce friction and improve morale, though they can increase instability by accelerating delivery. To balance this, value stream management (VSM) is becoming essential. By visualizing and measuring flow, VSM ensures AI-driven acceleration aligns with business priorities, not just raw throughput.

Augmenting vs. Evolving Workflows
AI-driven transformation takes two forms:

• Augmenting existing processes: streamlining reviews, strengthening data infrastructure, and modernizing compliance.
• Evolving new workflows: AI-native delivery pipelines with continuous testing, auto-detecting security systems, and hybrid human-AI collaboration models.

Most developer teams will need to do both, layering augmentation onto legacy systems while experimenting with AI-native patterns.

Developers in the AI-Native Era
The report also explores AI's impact on developer experience. Engineers using AI report higher "authentic pride" in their work, even if their sense of "meaningfulness" doesn't change. Prompt engineering is emerging as a core skill, redefining what it means to be effective in modern dev roles.

But risks are clear: with senior developers empowered to self-serve via AI, junior engineers may lose opportunities to build foundational skills. Organizations will need to treat developer growth as seriously as throughput—building intentional pathways for learning in AI-rich environments.

From Elite Practices to Universal Baseline
Earlier DORA reports focused on elite performers. The 2025 edition declares a new reality: AI is the baseline. The key question is no longer whether developers use AI—it's whether organizations have the systems, practices, and platforms in place to harness it effectively.

As the report concludes, AI adoption is inevitable, but AI transformation is optional. For developers, that means the challenge ahead isn't about adding one more tool—it's about reimagining how software itself is built.