The 2025 DORA Report: An engineering leadership perspective
Chris Westerhold, Thoughtworks’ Global Practice Director of Engineering Excellence, shares his insights on the 2025 DORA Report’s major research findings.
The 2025 DORA Report highlights a profound transformation as AI becomes embedded in software engineering, delivering notable gains in productivity, throughput and quality. Yet, these advancements have surfaced new challenges: increased delivery instability, the risk of speed overtaking long-term stability and fragmented toolchains.
From the Thoughtworks perspective, truly sustainable success with AI requires more than surface-level adoption. Organizations must look beyond immediate productivity boosts to embrace systems thinking, integrated platforms, actionable insights and a strong engineering foundation. These elements help manage the complexities and risks of accelerated change — unlocking lasting value from AI.
How is AI really impacting work?
AI is rapidly reshaping what’s possible in software engineering. Organizations are harnessing AI to accelerate delivery cycles, unlock new levels of productivity, and empower teams to focus on high-value innovation. The 2025 DORA report captures this momentum: companies adopting AI are realizing gains in individual effectiveness, throughput, quality and overall performance.
But these benefits come with important considerations. The report highlights that, alongside improved outcomes, some organizations are experiencing increased delivery instability and ongoing challenges with developer experience. This underscores a vital truth: AI doesn’t transform fundamentals on its own. Intentional action is required to remove friction, elevate engineering practices, and create a foundation where AI can drive measurable, lasting value.
The productivity paradox: More code, more problems?
At first glance, the metrics are impressive. Developers using AI assistants can generate code at remarkable speed. But what happens when that code is integrated into systems with poorly defined architecture, weak testing practices or tangled deployment pipelines? As the DORA Report suggests, the answer is often instability.
Traditional measures of productivity, such as lines of code or story points, become misleading in an AI-augmented environment. A developer might produce ten times the volume of code, but if that code introduces subtle bugs, security flaws, or architectural debt, the net effect can be negative. Teams then spend more time on rework, debugging and managing fragile systems.
This is where a new kind of waste emerges — AI engineering waste. As organizations adopt AI tools, they frequently encounter challenges such as:
- Prompt-response latency: Engineers spend valuable time waiting for AI models to generate responses, delaying workflows and breaking focus.
- Context loss: If AI systems lose track of conversations or project-specific context, developers must repeatedly re-explain issues, leading to frustration and wasted effort.
- AI toolchain fragmentation: Teams juggle multiple, disconnected AI tools and platforms, which leads to frequent context switching and increased cognitive load.
- Validation overhead: Thoroughly reviewing and validating AI-generated code for correctness, security, and coherence adds significant effort to the process.
These forms of waste erode efficiency and contribute to developer burnout. Instead of streamlining workflows, poorly integrated or misapplied AI adds extra layers of work and complexity. Developers can become overwhelmed by the very tools meant to help — spending more time managing tools and validating outputs, and less time on meaningful problem-solving and innovation.
If your organization already grapples with:
- Inefficient planning and delivery: AI will help you build the wrong thing, faster.
- Poorly structured teams: AI accelerates work within silos, increasing integration challenges.
- Weak developer experience: AI tools may become just another fragmented part of a confusing toolchain.
Higher coding throughput is here, putting pressure on the software engineering ecosystem
The modern engineer’s value is in prompt engineering, solution architecture and validating AI-generated outputs — not just in writing code. Without the right structure and processes, AI can turn speed into chaos.
Why systems thinking is essential for real AI impact
A key challenge facing engineering teams today is a lack of systems thinking — a holistic approach that considers how all parts of the organization work together. Adoption of AI tools can be accelerated and increased with systems thinking; a holistic interconnected approach improves developer experience and provides a springboard for reducing engineering cognitive load and improving outcomes.
To unlock AI’s real promise, organizations must integrate tools, workflows and organizational knowledge. Taking a systems thinking approach provides organizations implementing AI engineering uplift initiatives with the following improvements:
- Unified developer experiences: An experience that is focused on developers and leads to a reduction in context switching and has increased adoption of AI tools.
- Consistent AI patterns: Engineering organizations can operationalize common AI patterns based on engineering use cases that need a reduction in friction. These patterns can now scale across the organization with ease.
- Measurable acceleration across the entire SDLC: Impact of AI investments in the SDLC can now be measured beyond just adoption of tools. System thinking means that developer friction is accounted for and removed with the proper holistic implementations.
When tools, processes and knowledge are truly integrated, AI can start to genuinely alleviate these pain points. For example, new solutions such as CodeConcise enable developers to query codebases and receive real-time, context-aware answers — eliminating wasted time searching for information. Strong knowledge management systems offer a single source of truth, making it easier for engineers and AI to surface the right information at the right moment.
AI will only streamline these challenges when it is part of a unified, well-structured system. By investing in these foundations, organizations enable AI to automate routine tasks, surface valuable insights and create efficiencies across the software lifecycle. Without this approach, AI risks becoming yet another layer of noise — adding to burnout rather than relieving it.
Building the right foundation for AI success
How can organizations truly harness AI’s power without falling into these traps? The answer is a strong foundation of engineering excellence. This isn’t about chasing trends — it’s about systematically strengthening the core elements of a healthy engineering culture.
For more on measuring what matters in the age of AI, don’t miss the Thoughtworks “Pragmatism in Practice” podcast featuring Abi Noda, CEO and Co-founder at DX, and Chris Westerhold, as they discuss how top engineering organizations are turning AI hype into measurable value.
1. Operate as an optimization engine
Lasting progress demands more than just visibility. It requires a well-designed operating model that drives continuous improvement throughout the organization. That means setting up clear processes for transforming insights into action — embedding experimentation, measurement and course correction into daily work.
The process starts with evidence-based data: teams systematically gather and analyze both technical and business data to spot bottlenecks, inefficiencies and new opportunities. Instead of following intuition, they use data to form hypotheses and run rapid, targeted experiments. Small pilots let teams quickly learn what delivers genuine value with minimal risk.
Crucially, technical metrics are paired with business cases and financial modeling. This ensures that changes truly align with organizational priorities and create tangible returns. When new approaches prove effective, the operating model supports scaling improvements across teams.
By embracing this experiment-led, evidence-backed approach, organizations build a culture of fast feedback, rapid iteration and shared learning. Over time, this turns insights into everyday strategies for performance and adaptability.
2. Strengthen the core with platform engineering
Modern platform engineering treats infrastructure as a product — with an eye on both present robustness and future flexibility. Platforms must meet today’s business needs while staying ready for tomorrow’s technologies. The most effective systems provide developers with a “paved road” — self-service tools, streamlined deployment and built-in guardrails — freeing them to focus on business value rather than infrastructure headaches.
Crucially, these platforms must be adaptable and technology-agnostic. That way, your organization avoids being locked into inflexible workflows or solutions. Platform maturity enables rapid integration of new AI capabilities and practices, fostering a culture of continuous innovation. Being ready for change ensures teams can quickly adopt new advances — making adaptability a long-term advantage.
A reliable platform becomes the backbone for integrating and scaling AI capabilities.
3. Prioritize the developer experience
A world-class developer experience is now essential, not optional. That means a centralized hub for tools and documentation, simple workflows and meeting developers where they are — whether that’s the IDE, CLI or portal. Reducing friction and boosting self-sufficiency not only raises productivity but also helps retain top talent. When developers feel empowered, they can leverage AI as a creative partner.
However, simply automating broken or unnecessary processes can backfire. Using AI to generate volumes of new documentation — without addressing the core problems — can lead to overload, outdated knowledge and increased maintenance. The real goal is solving the issues that drive repetitive questions or confusion in the first place.
Modern AI tools create a shift away from static documentation. Developers can query the codebase and get real-time, context-aware answers — no more hunting through endless pages. For example, if an engineer needs the rationale for a design or usage of a component, AI can instantly provide tailored, up-to-date responses. This streamlines information sharing, reduces redundant work and frees up developer time for creative problem-solving.
From AI-augmented to AI-first
The journey with AI is evolutionary. Most organizations are currently in an “AI-augmented” phase, using tools to support existing processes. True transformation comes with the shift to an “AI-first” approach — where intelligent agents and automated workflows fully redefine how software is delivered.
This future is accessible only to those who build the right foundations now. Focusing on engineering insights, platform maturity, and developer experience creates conditions where AI thrives. At Thoughtworks, our AI-First Software Engineering Transformation offering supports organizations making this leap, maximizing AI value across the entire software delivery lifecycle.
The 2025 DORA Report is both a celebration of AI’s potential and a timely reminder: organizations that invest in strong teams, sound practices and ongoing improvement will realize the greatest benefits from AI.
Thoughtworks is a proud sponsor of the 2025 DORA Report. You can download the full report here.
Originally published at https://www.thoughtworks.com.
