Pilots are the practice laps. Going live is race day. Lessons from 16 finance leaders at Circuit Zandvoort.
October 25, 2026
Accounting and finance people are all about accuracy and trust, which is perhaps why AI is slower on the adoption curve in finance. I’ve spent much of my career focused on the financial close, first as an accountant and now leading product marketing for close and consolidation products. It’s where the instinct that makes finance good at its job can also make it hard to hand work over. The caution isn’t without justification: Gartner found that 62% of CFOs say fewer than a quarter of their AI initiatives deliver measurable benefits.
I recently joined my colleague Peter Van Craenenbroeck in hosting a Leadership in Finance Summit roundtable at Circuit Zandvoort that drew 16 CFOs and senior finance leaders from industries ranging from fast-growing scale-ups to large international organizations. The session focused on one question: what are you ready to hand over to AI?
There was plenty of optimism in the room, and some rightful scepticism. Both are healthy. In finance, scepticism isn’t resistance, it’s the job. The best conversations were between people who saw the potential and those who asked what happens when it gets something wrong.
Everyone in the room agreed AI can take on real work, whether in close, consolidation or reporting. Many were already well on their way, with adoption taking hold in their own teams. The more interesting conversation was about control. Finance leaders still sign the numbers, and handing over the work doesn’t hand over the accountability. So, control shouldn’t be the price of delegating. If anything, it’s what makes delegating possible.
Deploying AI and getting value from it are not the same thing. Deloitte’s 2026 Finance Trends research found that of the respondents who say their organisation has fully deployed AI (63%), only 21% believe it has delivered tangible value so far. The picture isn’t all gloomy though. KPMG’s 2026 AI in Finance survey found most finance leaders say their AI returns are meeting or exceeding expectations, which suggests the gap sits between teams that can show the value and teams that can’t yet.
Put simply, plenty of teams have posted a cracking qualifying lap, but keeping race pace once it’s live is the hard part. Everyone looks quick in practice.
If you don’t trust the numbers going in, you won’t trust what comes out, and you’ll end up double-checking everything anyway. The teams that sorted their data first seemed to get furthest. Deloitte’s CFO Signals survey of North American CFOs points the same way, with 52% naming data quality, access, and usability as a top transformation priority.
Before you hand a process to AI, you need to map it as it really runs, not as the manual describes it. That means the exceptions, the judgement calls, and the workaround someone built years ago that nobody wrote down. Nuance is where automated processes tend to trip.
There tends to be a dividing line in finance. In close and consolidation, accuracy is paramount, partly because of regulation. In planning and budgeting, there’s more room to let AI do the heavy lifting. But the line isn’t fixed. The better your data and your process, the further down it goes. It’s less a wall and more a pace car you can ask to speed up.
AI treated as a side project in one team tends to stall. It works best as a company-wide initiative with a clear owner, shared standards, and finance at the table from the start.
Adoption is a people journey as much as a technology one. Teams that make it safe to get things wrong in the practice laps tend to learn faster because the practice lap nobody is allowed to get wrong isn’t much of a practice lap at all.
KPMG Canada’s 2026 AI in Finance report argues that value is held back by fragmented ownership, pilots without a clear path to business value or scale, and incomplete measurement. So, starting small isn’t timid as long as each step has an owner and a measure. It’s how you earn the right to go further, which is why these five questions are worth asking first:
If the roundtable left me with one thought, it’s that the instinct that makes finance so good at its job — the insistence on accuracy and trust — may also be what gets AI over the line. Not by being cautious, but by being deliberate: trusted data, mapped processes, clear ownership, and room to learn. Perhaps control isn’t what slows AI down. Perhaps it’s what gets it to race day.
So where is your line today, and what would it take to move it?
