AI leaders lack rules & data to scale finance tools

Research commissioned by spend management firm Payhawk found that most finance organisations that consider themselves AI leaders still lack key operational foundations to scale the technology into core finance workflows.

A global survey of 1,520 finance and business leaders suggests AI experimentation is now common, but governance and data readiness continue to lag. Payhawk defined "AI leaders" as respondents who rated their AI maturity between 7 and 10 out of 10. That group included 405 respondents.

Among these self-identified AI leaders, only 26% reported having all the foundations the study linked to moving from adoption to day-to-day operational deployment. The remaining 74% said they were missing at least one requirement.

For finance teams, operational deployment means using AI in processes with high accountability, such as financial close, internal controls, approvals, exception handling, audit trails, and spend governance. The findings point to a gap between early adoption and the ability to embed AI in workflows that require consistent oversight and traceability.

Five requirements

The study outlined five operational requirements, which determine whether AI can scale in finance workflows: execution measures in place, minimum rules for AI use, skills and tools, committed budget, and data usable for AI analytics.

Within the AI leader group, respondents were most confident about skills and resourcing: 78% reported strong availability of AI skills and tools, and 69% said they had committed AI budgets. Execution measures were also relatively common, with 64% saying they were in place.

Governance and data preparation were weaker. Nearly a third of AI leaders (32%) said they had the skills but lacked minimum rules for safer use, while another 22% said they had implemented AI measures but still lacked minimum rules for consistent scaling. Data readiness also emerged as a constraint: 39% did not strongly agree that their data could effectively support AI-driven analytics.

The results suggest AI adoption in finance may be constrained less by skills and more by control frameworks and data quality. The study described this as "rules debt" and "data debt", with activity moving faster than governance standards and trusted datasets.

Operational gap

Payhawk framed the findings as evidence of a shift in what holds finance functions back as AI moves deeper into operational work. Many organisations reported investment intent and some governance activity, but scaling can stall when minimum rules are unclear or when systems struggle to reconcile AI outputs with trusted financial data.

That matters because finance processes often require clear accountability, reliable audit trails and consistent controls. Automation can reduce manual effort, but it also raises questions about how decisions are made, how exceptions are handled and how outputs map back to financial records. When AI is used for approvals, reporting and spend management, organisations also face internal compliance expectations and, in some cases, external scrutiny.

The survey covered senior professionals across roles and regions, including C-suite leaders, vice-presidents, directors and senior individual contributors. Functions represented included finance, accounting, sales, HR and procurement. Respondents came from industries including services, digital, manufacturing, healthcare, education, non-profit and B2C businesses. Company sizes ranged from 50 employees to more than 1,000.

The geographic scope included DACH, Spain, France, Benelux, the UK and Ireland, and the United States. Payhawk partnered with IResearch, using affirmative statements developed with finance and business leaders and conducting interviews across eight countries.

Payhawk is based in London and sells spend management software covering bills, cards, expenses, travel and procurement. It also offers a global money account that sits on top of an organisation's ERP system, serving mid-market and large enterprises in more than 32 countries.

The survey also suggests finance teams may be moving quickly to test AI tools without establishing consistent operating rules across the function. That approach can work in pilots, where the scope is limited and oversight is direct. It becomes harder when AI is expected to run repeatable, controlled workflows across teams, systems and geographies.

"In finance, AI only matters when you can delegate real work inside controlled workflows like approvals, reporting and audit trails," said Hristo Borisov, Payhawk CEO and Co-founder. "Our data shows the skills and experimentation are already there. What's missing is the operating stack, minimum rules and usable data that make AI accountable at scale."