Your finance function is full of AI. Is it deciding any better?

There is a moment, somewhere in the review of the board pack, when it lands on you that every number in front of you has passed through a model. The forecast, the variance commentary, the scenario table: faster than last year, more detailed, produced in a fraction of the time. And a quiet question sits underneath it all. Is any of this actually better, or is it simply more? You approve the pack because the deadline is real. The unease stays.

Here is the answer, plainly. Your finance function has already won the adoption argument, and that is no longer where the advantage lives. More than three-quarters of large finance teams now run AI through planning, reporting and commercial analysis. The separation between the organisations pulling ahead and the ones simply spending is not how much AI they use. It is whether they can trust what it decides, and prove it. That is the shift KPMG's 2026 finance study puts a number on, and it changes where a leader should spend attention this year.

Is my finance team actually deciding better, or just faster?

Both, for a specific subset. Broad AI use is now near-universal in large finance functions, and 71% of the leaders KPMG surveyed say it is meeting or exceeding their ROI expectations. The standout decision gains, though, do not spread evenly. They concentrate among organisations deploying agentic AI for finance (systems that plan and act across several steps rather than answering a single prompt): decision quality improves for around 70% of them, decision speed for 71%, and forecasting accuracy for 64%.

This is the part worth holding onto. Those agentic adopters separate from the rest by about 32 percentage points on average, widening to nearly 40% on forecast accuracy and return. Notice where the value lands: in judgement-heavy work, not the transactional automation most teams started with. The tool is broadly the same across the market. The distance between the leaders and the rest is what they built around it.

Why does trust decide whether the AI pays off?

Because a number you cannot stand behind is a number you cannot use. KPMG's sharpest finding is about evidence. Organisations that can produce audit evidence for how an AI output was reached report significant error reduction at 33%, against just 6% for those that cannot. Same category of tools. More than five times the improvement, decided by governance rather than by the model.

There is a quieter gap underneath. When the Bank of England and the Financial Conduct Authority surveyed UK financial firms in 2024, 75% already used AI, yet only about a third claimed to fully understand the technology they were running. Usage has outrun comprehension, and that gap is exactly where an unexamined decision hides. "Adoption is no longer the differentiator," as Nikki McAllen, KPMG's Global Head of Finance Advisory, puts it; the leaders are building the operating conditions around AI, governance, measurement, and a workforce equipped to act on what it produces. This is the same pattern that sits behind why most organisations struggle to turn AI into profit, and it rhymes with a wider trust deficit, the one that leaves whole markets using AI heavily while trusting it barely.

How do I build the conditions that make AI worth trusting?

Less exotic than it sounds, and it follows an order. Trust is manufactured deliberately, in the same way a clean audit trail is. Build it in these steps, because the order is where most finance AI programmes quietly lose the plot.

1. Reframe around the decision, not the task. Start from the two or three decisions you want to make better this year, forecasting, capital allocation, pricing, then point AI at those. Value follows the decision, not the volume of output.
2. Make governance the enabler, not the brake. Insist every material AI output can show its working. Audit evidence is what turns a fast number into a number you will sign, and it is the single sharpest lever in the KPMG data.
3. Embed measurement from day one. Decide up front how you will know an AI-assisted decision was a better decision. Cadence, accuracy, error rate. What gets measured is what earns trust.
4. Transform the workforce, do not simply add the tool. Only 38% of organisations are upskilling their existing finance teams and 28% are hiring for different skillsets. The tool arrives in a day; the fluency to act on it is the slower, decisive build.
5. Fix the data foundation. 36% of finance leaders name data quality and integration as their biggest opportunity, which is a polite way of saying it is their biggest constraint. Better inputs, better decisions.

What does this change for me as a leader this quarter?

It moves the question you carry into the finance leadership meeting. Not "are we using enough AI", which you have answered, but "can we stand behind what it tells us, and are we deciding better because of it". That is a governance question and a people question, and both sit with you rather than with the vendor. The boards that are ahead here are the ones already asking the sharper questions about oversight and readiness, not the ones buying more licences.

There is a longer arc behind this shift. Phase One, the Age of Effort: work hard, get a little more, linear growth. Phase Two, the Age of Scale: build once, sell to many, exponential growth. Phase Three, the Age of Acceleration: output decoupled from human effort almost entirely, the phase AI unlocks. Finance already has one foot in Phase Three. The bottleneck is no longer the technology; it is the governance, the measurement, and the judgement of the people deciding what to trust. That is a leadership choice, and it is yours to make this quarter.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.