The State of AI-Assisted Software Development in 2025: Key Insights from DORA's Latest Research
This is a summary of DORA’s 2025 State of AI-Assisted Software Development report, highlighting key findings and practical advice for technology leaders navigating the AI transformation. It is based on survey data from nearly 5,000 technology professionals worldwide. This blog post provides an overview of the main insights, including the role of AI as an amplifier, the benefits and challenges of AI adoption, the DORA AI Capabilities Model, team profiles, and actionable recommendations for leaders.
The software development landscape has undergone a seismic shift. In just three years, AI adoption has evolved from a curiosity to a near-universal reality. According to DORA’s latest research involving nearly 5,000 technology professionals worldwide, 90% of developers now use AI as part of their work—a 14% increase from 2024.
AI is an Amplifier, Not a Magic Solution
The central insight from DORA’s 2025 research is striking: AI’s primary role in software development is that of an amplifier. It magnifies the strengths of high-performing organizations and the dysfunctions of struggling ones.
Think of it this way: If your organization has solid foundational practices—quality platforms, clear workflows, aligned teams—AI will accelerate your success. But if your systems are chaotic, AI will simply help you produce chaos faster.
The Good News: AI is Delivering Real Benefits
The research shows encouraging trends:
- 80% of developers report increased productivity from AI adoption
- 59% observe improved code quality
- AI now improves software delivery throughput—a key shift from 2024
- Developers report spending more time on valuable, meaningful work
Unlike last year’s findings, which showed AI actually decreasing throughput, teams have adapted. The tools have improved, people have learned how to use them effectively, and organizations have begun restructuring their workflows to accommodate AI-accelerated development.
The Challenge: Instability Persists
However, the picture isn’t entirely rosy. Despite productivity gains, AI adoption still increases delivery instability. This suggests that while teams are moving faster, their underlying systems haven’t fully evolved to safely manage AI-accelerated development.
The research also found that AI has had no measurable impact on reducing burnout or friction—two persistent challenges that require systemic, cultural solutions rather than technological ones.
The DORA AI Capabilities Model: Your Roadmap to Success
Perhaps the most valuable contribution of this year’s research is the introduction of the DORA AI Capabilities Model—seven foundational practices proven to amplify AI’s positive impact:
1. Clear and Communicated AI Stance
Organizations need explicit policies about AI use. Ambiguity creates both overcautious developers who underutilize AI and reckless ones who overstep boundaries. A clear AI policy amplifies AI’s benefits for individual effectiveness, organizational performance, and even reduces friction.
2. Healthy Data Ecosystems
AI is only as good as the data it trains on. Organizations with high-quality, accessible, and unified internal data see dramatically amplified returns from AI adoption.
3. AI-Accessible Internal Data
When AI tools can access internal documentation, codebases, and company-specific context, they become far more effective. This capability significantly amplifies improvements in both individual effectiveness and code quality.
4. Strong Version Control Practices
Frequent commits and the ability to quickly rollback changes become even more critical in an AI-accelerated environment. These practices amplify AI’s positive impact on individual and team performance.
5. Working in Small Batches
Teams that break work into manageable units see amplified benefits from AI, particularly in product performance and reduced friction—even if individual effectiveness gains are slightly lower.
6. User-Centric Focus
This is perhaps the most critical capability. Teams with a strong user focus see amplified performance benefits from AI. Conversely, without a user-centric focus, AI adoption can actually harm team performance.
7. Quality Internal Platforms
A robust internal platform acts as a force multiplier for AI, ensuring that individual productivity gains translate into organizational-level improvements. 90% of organizations have now adopted platform engineering, making it foundational rather than optional.
The Trust Paradox
Interestingly, while AI adoption is nearly universal, 30% of developers report little to no trust in AI-generated code. This healthy skepticism isn’t a problem—it’s a sign of mature adoption. The most successful teams embrace a “trust but verify” approach, using AI to accelerate while maintaining critical validation skills.
Seven Team Profiles: Where Does Your Team Fit?
The research identified seven distinct team archetypes, each with unique performance characteristics:
- Harmonious High-Achievers (20%) - The gold standard: high performance, low burnout, stable systems
- Pragmatic Performers (20%) - Consistently deliver with speed and stability, though engagement could be higher
- Stable and Methodical (15%) - High quality at a deliberate pace in a sustainable environment
- Constrained by Process (17%) - Stable systems but inefficient processes create burnout
- Legacy Bottleneck (11%) - Trapped in reactive mode by unstable systems
- Foundational Challenges (10%) - Struggling across performance, environment, and outcomes
- High Impact, Low Cadence (7%) - Strong results but with high instability and demanding environment
Understanding your team’s profile helps identify the right interventions rather than applying generic solutions.
Practical Advice for Technology Leaders
Based on these findings, here’s what successful organizations are doing:
Treat AI adoption as organizational transformation, not just tool deployment. The greatest returns come from investing in foundational systems: your internal platform, data ecosystem, and core engineering disciplines.
Shift training from usage to critical evaluation. Focus on teaching teams how to guide, evaluate, and validate AI-generated work rather than simply encouraging more AI use.
Use Value Stream Management as a force multiplier. VSM ensures that local productivity gains from AI translate into meaningful organizational improvements rather than creating downstream chaos.
Prioritize your platform. A poor developer experience and fragmented tooling will hamper even the best AI strategy. Organizations treating their platform as an internal product see significantly greater returns.
The Bottom Line
AI is transforming software development, but its impact is fundamentally shaped by the environment in which it operates. Organizations that invest in the seven AI capabilities—treating AI adoption as a systems problem rather than a tools problem—will see AI amplify their strengths.
Those that simply deploy AI tools without addressing foundational practices risk amplifying their weaknesses instead.
The future belongs not to organizations that adopt AI first, but to those that create the conditions for AI to thrive.
This blog post appeared first on jayvilalta.com.
The full 2025 State of AI-Assisted Software Development report is available at dora.dev. Join the DORA Community at dora.community to discuss these findings and learn from peers navigating the AI transformation.