The 2025 State of AI-Assisted Software Development: Key Insights from DORA

The following is a summary of key insights from the 2025 DORA “State of AI-Assisted Software Development” report.

In 2025, the question for technology leaders isn’t whether to use AI—it’s how to unlock its value. The latest DORA “State of AI-Assisted Software Development” report, based on nearly 5,000 professionals worldwide, reveals that AI has become a near-universal part of software development. But its impact depends on something deeper: the strength of your organization’s systems.

AI as an Amplifier

DORA’s research identifies a central truth—AI amplifies what already exists. High-performing organizations see their strengths magnified, while dysfunctional ones find their weaknesses exposed. The most successful adopters invest not just in tools but in foundational systems: strong internal platforms, clear workflows, and aligned teams.

AI Adoption Is Now Mainstream

90% of respondents use AI at work, up 14% from last year.
80% report increased productivity, though 30% still don’t fully trust AI-generated code.
Developers rely on AI for coding, debugging, documentation, and even communication tasks, spending a median of two hours per day with AI tools.

While AI use is widespread, a “trust-but-verify” mindset prevails—teams critically evaluate and validate AI’s output rather than accepting it blindly.

Seven Team Profiles

DORA identified seven archetypes that capture the balance of performance, well-being, and stability:

Foundational challenges – struggling to survive.
Legacy bottleneck – reactive, burdened by unstable systems.
Constrained by process – trapped in bureaucracy.
High impact, low cadence – producing great work, but too slowly.
Stable and methodical – deliberate, consistent progress.
Pragmatic performers – fast, stable, and effective.
Harmonious high-achievers – high-velocity, low-burnout excellence.

Notably, clusters 6 and 7—about 40% of teams—prove that speed and stability can coexist, disproving the old “move fast and break things” myth.

The DORA AI Capabilities Model

To guide adoption, DORA introduced a new framework identifying seven capabilities that enable AI success, including:

A clear, organization-wide AI policy
A healthy, accessible data ecosystem
A strong internal platform
A user-centric focus

These capabilities transform AI from a set of disconnected tools into a performance amplifier.

Platform Engineering and Value Stream Management

Platform engineering has reached 90% adoption, making internal platforms the backbone of AI-driven productivity. Teams that treat their platform as a product—improving developer experience—see the best outcomes.

Similarly, value stream management (VSM) acts as a force multiplier for AI, turning localized gains into measurable organizational improvements.

The AI Mirror

Ultimately, AI reflects the maturity of your engineering culture. Organizations with strong feedback loops, independent architectures, and a culture of learning benefit most. As Gene Kim writes in his foreword, “When AI accelerates development, our control systems—us—must speed up too.”

From Insight to Action

The takeaway from DORA’s 2025 report is clear: AI doesn’t fix broken systems—it magnifies them. Success comes from treating AI adoption as an organizational transformation, grounded in sound engineering practices, healthy data, and empowered teams.

For leaders, the next step is not to chase every new tool, but to strengthen the systems that will let AI’s potential become real, sustainable performance.