Artificial intelligence is taking on a growing role in tax due diligence, but according to a recent report from TAINA Technology, the next challenge for firms is not whether AI can complete the work, but who is responsible for managing it.
The analysis from Richard Kent, argues that tax due diligence has traditionally relied on human expertise built through knowledge of legislation, risk identification and professional judgement. While AI agents are now capable of reviewing documents, summarising information, flagging anomalies and drafting reports, their outputs still require oversight.
According to Kent, many firms are moving towards an operating model in which leaders manage people, and people manage AI agents. This represents a leadership challenge as much as a technological one, requiring professionals to combine people management, quality assurance and critical thinking.
Unlike traditional software, AI agents reason, interpret and prioritise information. As a result, they can produce conclusions that appear credible but are incorrect.
Kent argues that a strong grounding in first principles remains essential. Rather than simply knowing the answers, tax professionals need to understand how conclusions are reached by applying knowledge of transaction structures, tax concepts, legislative intent and commercial context.
Without that understanding, managers may fail to identify when an AI agent has overlooked a material risk, misunderstood a fact pattern or reached an incorrect conclusion. The analysis also suggests there is a risk of accepting AI-generated outputs because they appear authoritative.
As AI capabilities continue to develop, the role of human judgement does not diminish, according to the assessment. While AI agents may be able to identify dozens of potential tax issues during a due diligence review, professionals are still responsible for determining which findings are material, which represent false positives, what further evidence is required and how those findings should influence a transaction.
The analysis suggests this places new demands on firm leaders, whose responsibility extends beyond developing technical expertise and people management skills. Teams must also learn how to direct, challenge and evaluate AI-generated work by asking effective questions, verifying conclusions, understanding the limitations of AI outputs and recognising when human intervention is required.
To illustrate the point, the analysis compares AI agents to exceptionally fast graduates that can process large volumes of information and produce strong first drafts, but still require review, challenge and oversight from an experienced partner.
According to TAINA Technology's analysis, firms are likely to gain the greatest value from AI not simply by adopting the technology, but by developing the capability to manage it effectively.