Finance AI stuck in risky mid-maturity 'AI middle'

Payhawk has published research suggesting that AI adoption in finance has moved beyond the early stage, but remains uneven across organisations. Many teams are struggling to shift from experimentation to day-to-day use.

The survey of 1,520 senior finance and business leaders found that around half of organisations sit in a mid-maturity band for AI use in finance. They report active adoption but do not yet run AI as a core finance capability. Payhawk describes this as an "AI middle", where activity raises internal expectations while governance and operational consistency lag.

Payhawk's CFO AI Readiness Report asked respondents to rate their organisation's AI maturity on a 1-10 scale. Scores of 1-3 were classified as low maturity, 4-6 as mid, and 7-10 as high.

Nearly one-third of organisations rated themselves in the high-maturity range. Payhawk said this makes the "leader" label broad and in need of closer examination. The results point to a split market: a smaller group scaling AI, a large mid-tier trying to turn pilots into repeatable processes, and another group still at an early stage.

Execution risk

The findings highlight operational risks between early trials and scaled deployment, particularly in finance functions where governance requirements are more stringent than in other parts of the business. Audit demands, control frameworks, and accountability processes can prevent tools from moving into core workflows.

Payhawk said the mid-maturity segment is a pressure point for chief financial officers as they make automation spending decisions. AI budgets are often scrutinised against expected productivity gains, while finance leaders still need to maintain oversight of policies, controls and reporting.

Hristo Borisov, Payhawk CEO and Co-founder, said the greatest risk lies in partial adoption rather than early exploration.

"The real risk in finance AI isn't experimentation, it's getting stuck halfway," Borisov said.

Sector differences

AI maturity varies by sector and company size, according to the report. Tech organisations with more than 251 employees reported the highest maturity levels, with more than 70% rating themselves as highly mature. Among smaller organisations in regulated and "core-economy" sectors, defined as industries such as manufacturing, retail, logistics, energy and healthcare delivery-13.5% reported high maturity.

Large non-tech organisations clustered in the mid-maturity range. Many have started using AI but face challenges embedding it into core finance operations.

Organisational structure also correlates with higher self-reported maturity. Complex, multi-entity businesses were more likely to report stronger AI maturity, which the research links to the standardisation, shared services and centralised controls that often come with scale.

However, structure alone does not guarantee readiness. The report flags data consistency and alignment as issues that can slow progress even where governance models exist.

Leaders and laggards

Another conclusion is that organisations describing themselves as AI leaders may not be implementing AI in similar ways. Payhawk said headline maturity scores can mask differences in accountability, control design and operational discipline.

Some organisations report AI embedded into finance workflows with clear ownership. Others report rapid adoption without guardrails or investment plans that lack the foundations for repeatable deployment.

Payhawk argues that the limiting factor for maturity is less about the underlying models and more about making AI use stable and repeatable in financial control environments. It frames this as a question of building defensible processes around AI in areas that affect reporting, compliance and internal controls.

The survey was conducted with research firm IResearch and covered eight countries in Europe and the United States, including the UK and Ireland, France, Spain, Benelux, DACH and the EU. Respondents included C-suite leaders, vice-presidents, directors and senior individual contributors across finance, accounting, sales, human resources and procurement.

Payhawk, which sells spend management software, describes itself as a finance orchestration platform combining corporate cards, expense management, accounts payable and procure-to-pay. The company is headquartered in London, has offices across Europe and the United States, and operates in more than 32 countries.

"Many finance teams now have visible AI activity but lack the minimum structure needed to scale it safely under audit and control. The organisations that succeed won't be the loudest adopters, but the ones that make AI governable inside their finance operating model," Borisov said.