AI, innovation and R&D tax relief: drawing the line between adoption and advancement

AI, automation and data-led innovation are reshaping how businesses develop products, improve services and scale operations. But as AI becomes more widely adopted, it also creates a practical challenge for R&D tax relief: not every AI-led project will qualify.

The UK’s R&D tax regime has also changed. The merged R&D expenditure credit scheme and enhanced R&D intensive support apply for accounting periods beginning on or after 1 April 2024, replacing the previous SME and RDEC regimes for most claimants. The reforms are intended to simplify the landscape, but they also put greater emphasis on clear evidence, accurate cost treatment, claim notification requirements and robust compliance.

Where AI innovation can qualify

R&D tax relief still depends on whether a project seeks an advance in science or technology and involves resolving genuine scientific or technological uncertainty. Using AI, machine learning, automation or advanced analytics does not qualify on its own.

Drawing the line between adoption and advancement

For AI and software-led businesses, the key question is where the qualifying R&D sits. The advance may not be in the AI tool itself, but in the system architecture, integration approach, data engineering or technical constraints that need to be overcome.

Routine use of existing AI tools, including generative AI to improve productivity, is unlikely to qualify. However, developing a bespoke AI model or platform may involve qualifying activity where the team must overcome uncertainty around data quality, model performance, scalability, latency, integration or security.

Define the qualifying project clearly

AI often forms part of a wider commercial programme, so project boundaries need to be clear. Businesses should identify where the technological uncertainty arises, whether the advance sits in the AI component or the wider system, and how the project can be described in technical rather than commercial terms.

This helps distinguish qualifying R&D from routine development, implementation, optimisation and commercial rollout.

Key areas to consider under the new regime

AI and software-led businesses should pay particular attention to:

• AI-assisted development tools, where the claim should focus on the underlying technological uncertainty, potentially in computer science, rather than the use of the tool itself.
• Contracted-out R&D, including contractual responsibility, technical decision-making and who intended the R&D to be undertaken.
• Cloud computing and data costs, which may be eligible where they directly support qualifying R&D but need clear apportionment if also used commercially.
• Overseas R&D activity, where relief is now more restricted and cost or labour availability alone will not usually be enough.
• Claim notification requirements, particularly for first-time claimants or businesses that have not claimed recently.

Why evidence and compliance matter more than ever

HMRC scrutiny remains high, so broad descriptions of innovation are unlikely to be enough. Businesses need to show where uncertainty arose, what work was undertaken, why the challenge could not be readily resolved by a competent professional in the field, and how qualifying activity and costs have been separated.

Claims are strongest when considered early, supported by technical documentation, clear project scopes, evidence of experimentation and a consistent approach to cost apportionment.

Other innovation incentives may also be relevant

R&D tax relief should not always be considered in isolation. Depending on the business and the nature of the innovation, other incentives may also be relevant, including Patent Box, Research & Development Allowances and investment reliefs such as EIS or VCT. Aligning these incentives with a coherent innovation strategy can strengthen both tax outcomes and investment positioning.

What this means for AI and software-led businesses

AI may accelerate innovation, but it does not remove the need for structured thinking, clear documentation and a disciplined approach to R&D tax relief. Businesses best placed to benefit will be those that focus less on the tools they use and more on the technical challenges they are solving.

Those that can evidence the uncertainties faced, the work undertaken and the advances sought will be in a stronger position to support claims, withstand HMRC scrutiny and make better use of wider innovation incentives.

We are here to help

Innovation deserves the right support. At Azets, our R&D tax specialists work with AI, software and data-driven businesses to assess whether activity is genuinely qualifying, identify the right claim position under the new regime, strengthen technical evidence and prepare robust, compliant claims. If you are investing in AI, automation or data-led innovation and want to understand how the latest R&D tax rules could affect your business, get in touch with our specialist team.