The AI readiness gap that's holding AP back

Ask most finance leaders whether their organization is ready for AI, and they'll say yes. They're not wrong, exactly. A new Forrester Consulting Opportunity Snapshot commissioned by Basware, From Experimentation to Execution: Governed Autonomy for AI in Accounts Payable, found that most finance teams already have the AI guidelines, leadership backing, and risk frameworks in place. Strategy is not the problem.

So what is? Execution. Far fewer of those same organizations have built an operating model that can run AI day to day, or the team behind it to keep it that way. That's the real AI readiness gap in accounts payable, and it has nothing to do with ambition.

It matters because the pressure on AP isn't letting up. 65% of finance teams say adjusting to new financial regulation needs major or urgent improvement right now, and that pressure lands on whichever team owns AI in AP. Strategy alone won't carry that weight.

Over 60% of finance teams already have an AI strategy

Most finance leaders have already done the hard thinking. 67% have deployed AI in at least one targeted accounts payable use case, and 76% plan to increase AI investment further. Fraud detection and risk management lead the priority list, followed closely by reporting, analytics, and dashboarding: the two areas where finance teams feel most confident putting AI to work.

Add the strategic backing, and the picture looks even stronger. Over 60% of organizations have documented AI principles, executive sponsorship, and risk frameworks already in place. That's a genuinely solid foundation. It's also not where AI in accounts payable gets stuck.

None of this is about buying more AI tools. It's about proving that the AI already running fraud checks and invoice coding is accountable to someone, every single day, not just at go-live.

That distinction matters most for organizations running AP across several ERPs and countries, where a strategy signed off centrally still has to translate into consistent practice in every entity. Without operational discipline, even the best of strategies is no use. And this is exactly where multi-entity finance teams tend to feel the strain first.

So where does the AI readiness gap actually live?

Strategy tells an AI program what it's allowed to do. But the operating model is what runs it, day after day, invoice after invoice. This is where the numbers drop fast: only 46% of finance teams say they've struck an effective balance between governance and innovation, and just 39% have a center of excellence in place.

Without that structure, AI stays a collection of pilots instead of a system. Someone owns the fraud detection model. Someone else owns the invoice coding tool. No one owns the operating model itself. No one decides when to widen AI's remit, or proves to an auditor why a decision was made. That's exactly the audience this data speaks to loudest: teams running AP across multiple ERPs and geographies, where inconsistent governance compounds fast.

64% of finance leaders already sense this. They say they'd rather have AI that's stable and compliant than AI that's simply more capable. That's not caution for its own sake. It's a signal that control, not capability, is what's actually missing. The technology choices back this up: machine learning, robotic process automation, and predictive analytics run most of AI in AP today. Agentic AI and generative AI, the two technologies getting the most attention elsewhere, are still the least used.

Regulation raises the stakes further. With mandates like ViDA, France's B2B rules, and Germany's e-invoicing requirements arriving in quick succession, an AI decision that can't be explained to a tax authority isn't just an internal weakness. It's a compliance risk. An operating model gap that would have been a productivity issue five years ago is now a regulatory one.

Why does governing AI beat deploying more of it?

The instinct when a program stalls may be to add more AI: another model, another use case, another vendor demo. But this research suggests the opposite move works better. Organizations that treat the operating model as the priority, not an afterthought, are the ones that turn strategy into results.

That's the idea behind Governed Autonomy: AI that acts, a human who decides how far it goes, and an audit trail that defends every decision it makes. It isn't a feature to switch on. It's the operating model itself: the structure that lets an AP team increasingly hand more work to an AI, while building trust over time, just as one would with a new team member. This approach empowers teams to be confident the AI is completing tasks correctly, every time.

In practice, closing the AI readiness gap looks specific rather than dramatic: a named owner for each AI use case, a standing review of where AI's decision rights should expand or contract, and an audit trail that's checked routinely, not assembled after the fact when a regulator asks. None of this is glamorous. All of it is what separates a program that scales from one that stalls at the pilot stage.

The more useful question to bring to the next leadership review isn't “how much AI have we deployed.” It's “who can tell me, entity by entity, how each AI decision was governed.” The first invites a status update. The second invites an operating model.

With 68% of finance leaders demanding demonstrable ROI before they'll back further AI investment, that structure does something else too. It turns “we think this is working” into “here's what it delivered, and here's how we know.” That's the difference between an AI pilot and an AI operating model.

The gap between AI ambition and AI execution in accounts payable is real, and it's closeable. It starts with structure, not the next AI purchase, and not another AI vendor demo. For the teams that get this right, the payoff isn't just efficiency. It's a defensible answer the next time a board, auditor, or regulator asks how AI decisions get made.

Ready to see the full picture? Read the full study, From Experimentation to Execution: Governed Autonomy for AI in Accounts Payable, to learn what's driving the gap between AI strategy and execution in AP, and what it takes to close it.