Report finds AI coding tool impact likely overstated

However, companies which actually track those improvements consistently report more conservative gains than those operating on impression alone, suggesting that leadership confidence in AI is, for many organisations, running ahead of the evidence.

Measured productivity gains attributed to AI coding tools sit at around 20% on average across the firms that quantify them, with some teams reporting concrete wins such as project timelines halved and sprint output up 20% to 25% after adopting modern coding assistants. The report makes clear these are real gains worth pursuing. The challenge is that, in the absence of measurement, perceived gains tend to drift higher than the figures organisations would actually arrive at if they tracked them. This is likely because engineers feel much faster with AI than they truly are and may find that their mental burden is lower when using it. In reality, code rewrite rates can creep up and engineers can spend more time on reviews and rearchitecting poorly designed code, something which qualitative feedback and positive bias can obfuscate.

That gap matters because investment is already large and accelerating. 91% of businesses are now investing in AI tools within engineering, yet only 22% have a formal documented AI strategy and 83% are not tracking any metrics to quantify AI’s impact. Without that data, leaders cannot reliably tell which parts of the software development lifecycle AI is genuinely accelerating, where it is quietly creating rework, or which teams and use cases deserve more (or less) investment. The result is one of the largest line items in engineering budgets being managed on anecdote rather than evidence.

“AI is producing genuine productivity gains in UK engineering teams, the data on that is clear. But the gains people feel are often larger than the gains they would actually find if they measured them. The leaders who will get the most out of AI over the next 12 months are the ones who measure what is happening in their teams and then double down where the evidence is strongest, not the ones with the most tools or most extreme adoption,” said Mike Daniel, HI Technology & Innovation.

Why this matters now

AI is firmly established in UK software engineering. The question for technology leaders in 2026 is no longer whether to invest, but how to invest well. The report argues that strategy and measurement are what separate organisations that are already capturing AI’s gains from those that are spending without a clear picture of return. The companies which pair their AI investment with a written strategy and a small number of trustworthy productivity and quality metrics are best placed to direct AI where it delivers most, accelerate teams that are already winning with it, and avoid expanding it into areas where the evidence does not yet support it.

Key findings

• AI is delivering real gains, around 20% when measured. Firms tracking AI’s effect on engineering productivity report meaningful improvements on average, with some seeing project timelines and sprint cycles materially reduced.
• Most leaders are flying blind. 83% of UK businesses are not tracking any metrics to measure AI’s impact on engineering productivity.
• Measured gains are smaller than perceived ones. Companies that measure AI effectiveness consistently report more conservative velocity gains than those that do not, suggesting that unmeasured estimates tend to overstate AI’s impact.
• Strategy is the missing ingredient. 91% of businesses are investing in AI tools, but only 22% have a formal documented AI strategy for engineering.
• Depth beats breadth. Teams that focus AI on coding and debugging report roughly twice the velocity gain of teams that spread AI across four or more stages of the software development lifecycle.
• Quality is the most cited concern. Only 57% of engineering leaders are pleased by AI output quality. However, 79% report no increase in time spent on code review, quality assurance or testing despite the surge in AI-generated code volume.
• Shadow AI fills the strategy gap. Around half of engineers (52%) use AI tools even when their employer has not formally invested in them, raising IP, security and compliance risk.

The implication for technology leaders
 
The report concludes that AI is producing real value in UK engineering teams, and that this value is likely to compound over the next 12 months for the organisations best set up to capture it. Practically, that means defining a written AI strategy, instrumenting a small number of trustworthy productivity and quality metrics, and using those metrics to decide where to expand AI use and where to refine it. The opportunity is not to dial AI back, but to find the parts of the engineering process where its impact is clearest and invest there with confidence.

About the report

AI in Software Engineering: Making Sense of the Noise synthesises insights from more than 100 interviews with CTOs, CISOs, VPs of Engineering and senior engineers, alongside published academic and industry research. It is intended as a guiding resource for technology leaders defining, deploying and measuring AI within their engineering organisations.

The full report can be downloaded free of charge at:

hitechnology.io/lp/ai-in-software-engineering-report/