AI in Finance (May 2026)

In late 2025, the AICPA and CIMA asked 1,446 senior finance leaders whether they expected AI to be the most transformative trend in accounting and finance over the next two years. Eighty-eight per cent said yes. Eight per cent said their own organisation was "very well prepared" to manage that trend. This brief is for the 92%.

It is not a deployment guide. It is a guide to thinking about AI in finance clearly enough to govern it: what the term actually means, what the published evidence does and does not support, where the genuine risks sit, and what a finance leader can usefully ask of their team in the next ninety days.

What's inside

• A four-part taxonomy of classical ML, document AI, generative AI, and agentic AI, and why most failed projects begin with teams that did not know which of the four they had bought.
• A maturity map of finance use cases: where to deploy now (AP, treasury forecasting), where to pilot (close, FP&A, tax), and where the evidence still says experiment, do not commit (agentic execution across the function).
• What AI still doesn't do well, grounded in the FailSafeQA, FAITH, PHANTOM, FinanceQA, FinBen and AuditBench benchmarks, and why "ninety per cent accurate" is not a model you can deploy unsupervised against the general ledger.
• The five risk surfaces a CFO owns, the build-versus-buy decision and the thin-wrapper trap, and six questions to put to your team this quarter.

Leave your details below and we'll send you a copy by email.