Financial AI is establishing itself as one of the main levers of transformation in financial departments. However, the advancement of financial AI is not always accompanied by the necessary foundations to guarantee consistent, governed and secure use. In fact, almost half (45%) of companies considered leaders in financial AI still do not have minimum standards governing its application within key workflows.
This is reflected in the ‘CFO AI Readiness Report’ study, prepared by Payhawk, a global solution for corporate expense management, in collaboration with IResearch. The report analyzes the gaps that determine the ability of organizations to scale financial AI within economic and accounting processes, based on a survey of 1,520 financial managers and directors of companies from different sectors, sizes and regions. Among them, “leaders in financial AI” companies are those organizations that rated their AI maturity with a score of 7 to 10 out of 10.
“There is a perception that maturity in financial AI advances linearly, but the reality is more complex. Even among the most advanced organizations, preparation is uneven and is conditioned by different barriers. In this context, the main limit is not the capacity of financial AI technology, but to what extent organizations can govern it. That is, if they are able to justify it, track it and audit its use within financial processes,” explains Laura Gámiz, director in Spain at Payhawk.
The conditions that determine whether financial AI can scale in finance
According to the Payhawk report, for financial AI to move from initial adoption to fully operational use within financial workflows, organizations must meet five basic conditions: have execution metrics in place, have minimum standards governing the use of financial AI, have teams with the necessary capabilities and tools, allocate specific budgets, and have data to support analysis powered by financial AI.
However, achieving this level of preparation remains rare. Only 26% of organizations considered leaders in financial AI meet all five requirements, highlighting that even among the most advanced companies, adoption of financial AI does not always translate into fully-established operational capability.
An adoption of financial AI with different levels of readiness
Based on these criteria, the report segments leading organizations in financial AI into six operational postures based on their degree of development in the five requirements. Only the so-called “scaled adopters” (26.9%) have a fully consolidated model and meet all the necessary requirements to operate financial AI effectively. Next to them are the “gradual improvers” (17.5%), who present a partial and balanced preparation in the different dimensions of financial AI, without any one standing out particularly.
The rest respond to more unequal levels of development in the adoption of financial AI. “Execution-driven implementers” (16.0%) have good operational capacity but lack minimum standards for financial AI, while “agents first, control second” profiles (14.1%) have enthusiasm for financial AI experimentation outweighing governance, with no adequate frameworks or sufficient preparation in place. “Governance-first scalers” (13.8%) are also identified, with stronger control structures in financial AI, but with limitations in data quality. Finally, there are “control-first planners” (11.6%), who have relatively developed capabilities, budget and data for financial AI, but without implemented execution metrics, which shows that preparation does not always translate into operational deployment.
The main obstacles to scaling financial AI
The research highlights a clear imbalance: while 78% of financial AI leaders report having strong skills and tools, only 55% have minimum governance standards for financial AI, the lowest-rated readiness factor.
These operational approaches respond to two structural gaps: “standards debt” and “data debt” in the context of financial AI. The first occurs when organizations deploy financial AI solutions faster than they develop their governance frameworks, resulting in systems that are difficult to audit, explain, or securely integrate into financial workflows that require compliance and control.
For its part, “data debt” in financial AI appears when execution capabilities and control mechanisms are in place, but the underlying data is inconsistent, incomplete or fragmented. In these cases, organizations can manage the use of financial AI, but not fully trust its results at scale. This explains why some companies, despite having strong financial AI governance, fail to scale it in critical financial operations, especially in complex and highly regulated environments.
Only 55% have minimum governance standards for financial AI, this being the lowest rated readiness factor
Ultimately, the main challenge in adopting financial AI is accurately diagnosing each organization’s maturity point. An incorrect diagnosis can lead to inefficient investments, such as increasing financial AI capacity when the real obstacle is a lack of governance, or strengthening regulatory frameworks when the problem lies in data quality.
“Scaling financial AI remains complex because organizations advance unevenly in the capabilities that make it possible,” concludes Gámiz. “Many companies are investing in more financial AI when the real bottleneck is in standards or data. Scaling financial AI is, at its core, a coordination exercise that involves aligning governance, data and accountability across financial processes. Those organizations that only address a portion of these requirements will continue to encounter limitations and fail to move beyond certain use cases.”
