The smart tax function: leveraging generative AI 

International Tax Review is part of Legal Benchmarking Limited, 1-2 Paris Garden, London, SE1 8ND

Copyright © Legal Benchmarking Limited and its affiliated companies 2025

Accessibility | Terms of Use | Privacy Policy | Modern Slavery Statement

The smart tax function: leveraging generative AI 

Sponsored by

sponsored-firms-kpmg.png
Robot pressing screen.jpg

Niklas Elofsson and Vsevolod Konyshev of KPMG Sweden explore how tax functions can unlock demonstrable efficiency gains by using generative AI tools in a budget-constrained transformation landscape

Strategic use of generative AI in tax: delivering measurable returns

Tax functions can strategically leverage their organisation’s existing generative AI solutions to enhance efficiency and build a compelling case for future investments. This is especially relevant in today’s challenging economic climate, where securing budgets for new tax transformation initiatives is difficult. By concentrating on maximising the value of existing generative AI solutions and refining processes, tax functions can demonstrate tangible efficiency gains from, and improved utilisation of, existing investments, thereby building a strong foundation for subsequent transformation projects.

Strategic implementation of generative AI for routine tasks frees up tax professionals for higher-value work such as strategic planning, business partnering, and value creation. This allows tax teams to apply their expertise to complex analysis, nuanced interpretation, and strategic advisory roles, thereby enhancing their ability to deliver added value.

To effectively leverage existing generative AI solutions, it is important to understand its specific strengths: it excels at flexible, interpretive tasks with unstructured data, whereas traditional automation handles precise, repetitive operations. Understanding this fundamental difference is key for effective resource allocation.

Foundations for effective generative AI adoption

Successfully leveraging the value of initial investment in generative AI technology requires utilising a flexible and actionable strategy, often informed by internal or external experts. This strategy must define appropriate use cases and specific application opportunities with practical examples, providing a foundation for experimentation and deployment. Alongside this strategic foundation, success also requires a robust tech platform and cultivating a culture of innovation.

Leading tax functions that successfully integrate generative AI strategically use robust, evolving technology platforms. These platforms ensure access to a diverse range of leading models from various providers and incorporate essential supporting capabilities, facilitating early adoption of the latest advancements. Importantly, these leading functions select specific generative AI solutions based on technical capability, user-friendliness, scalability, and integration potential.

Maximising the impact of generative AI adoption hinges on successfully fostering a culture of innovation built on practical applications and standard output verification. This involves incentivising experimentation, where lessons from both successes and failures are valued. Continuous learning programmes – progressing from basics to advanced applications, supported by ongoing knowledge sharing through regular follow-ups and forums for teams to discuss insights and challenges – underpin this culture.

Another key feature of many organisations successfully harnessing the power of generative AI is the ability to leverage motivated champions within the organisations who understand both technology and organisational goals. Creating an environment where these champions can drive adoption and transparent communication about AI capabilities helps manage expectations and guide practical usage.

A flexible, iterative approach for generative AI implementation

To demonstrate tangible improvements, a flexible, iterative approach to generative AI implementation must be adopted. In practice, this typically means beginning by identifying ‘time thieves’ – processes consuming disproportionate manual effort or involving repetitive, high-volume tasks. These become ideal candidates for initial generative AI pilots. Once a target process is selected, its steps are mapped to pinpoint exactly where generative AI can deliver the most impact, serving as a focused testing ground without disrupting entire workflows.

The authors’ observations from working with successful adopters show that small pilot projects are crucial to successful generative AI endeavours. They provide critical insights, reveal generative AI’s practical strengths and limitations, and allow for testing capabilities before scaling.

This iterative process of piloting process improvement projects, combined with the consistent and early capture of lessons learnt, is vital not just for the immediate use case but for shaping the broader adoption strategy and future implementations. These learnings enable strategic decisions; for example, if a current generative AI solution falls short, leading functions maintain the flexibility to explore advanced versions or even non-AI alternatives, ensuring their overall approach accommodates emerging capabilities and evolving organisational needs.

Cross-functional collaboration, often including internal or external specialists where beneficial, is essential for effective generative AI adoption. Drawing on diverse perspectives and proven strategies accelerates learning, avoids duplicating efforts, and ensures that the entire tax function benefits. This collaborative approach helps to maximise value provided by existing technology. This approach also facilitates the identification of future automation opportunities and the development of well-founded cases for further investments.

The power of strategic integration of generative AI

Through pilot projects and continuous learnings, leading tax functions gain a deep understanding of process steps in which the deployment of generative AI provides the biggest efficiency gains, while also learning the limitations. Based on this information, they maximise value for investment by strategically integrating generative AI workflows with other established technologies, rather than using it as a standalone solution.

These tax functions integrate generative AI’s unique strengths into workflows that rely on proven tools already in use for precision tasks such as data validation, complex calculations, or rule-based automation, offsetting the weaknesses and boosting the strengths of both. This layered approach ensures generative AI’s outputs are refined, verified, and operate within controlled, auditable frameworks, thereby creating trustworthy and highly effective end-to-end tax processes.

Key takeaways on tax functions’ use of generative AI

Ultimately, effective generative AI implementation in tax is accomplished by utilising a flexible, methodical, and strategic approach that fosters a culture of innovation, emphasises practical application, and embraces learning-by-doing. Leading tax functions demonstrate how leveraging generative AI tools can achieve measurable gains with comparatively small investments.

This balanced methodology – strategically targeted, realistic, and adaptable – empowers tax professionals to show tangible returns, strengthening stakeholder confidence and paving the way for broader transformation. By concentrating on focused, iterative pilots that quickly deliver visible impact, tax leaders secure immediate wins while positioning their departments for ongoing evolution in an increasingly complex global environment.

more across site & shared bottom lb ros

More from across our site

Saffery cautioned that documentation requirements in new government proposals must be limited if medium-sized companies are not exempted from TP
The global minimum tax deal is not viable without US participation, Friedrich Merz has argued
Section 899 of the ‘one big beautiful’ bill would have spelled disaster for many international investors into the US, but following its shelving, attention turns to the fate of the OECD’s pillars
DLA Piper’s co-head of tax for the US and Latin America tells ITR about her fervent belief in equal access to the law, loving yoga, and paternal inspirations
Tax expert Craig Hillier agrees with the comparison of pillar two to using a sledgehammer to crack a nut
The amount is reported to be up 57% from the £5.6bn that the UK tax agency believes was underpaid in the previous year
The US president also unveiled a new 50% levy on copper imports; in other news, a UK wealth tax proposal has been criticised by the Institute for Fiscal Studies
Wim Wuyts, who had been head of the specialist tax network since 2017, is moving on to a new role with WTS’s Belgian member firm
MNEs are increasingly using algorithmic tools in TP. Sahasranshu Dash argues that data ethics should therefore plug directly into the TP design process
The Institute of Chartered Accountants in England and Wales also queried whether HMRC resources could be better spent scrutinising larger entities
Gift this article