Burning Issues #1: AI is reshaping Finance

How C-suite and CFO are deploying artificial intelligence

At C-suite level, AI is often framed as a strategic capability: a tool to anticipate and support better decision-making. However, turning that promise into immediate impact proves challenging. Most organizations seem to focus on efficiency gains: how to free up capacity for higher-value activities, and, increasingly, how to deliver the same output with fewer full-time equivalents.

1. Benefits for the CFO

For CFOs, this is not a passing wave of innovation. AI is reshaping the core of the Finance function. Where Finance once focused on the past (actuals versus prior year and budget) it is now looking decisively forward, with greater emphasis on rolling forecasts and scenario planning. That shift is both necessary and overdue.

Only four to five years ago, Robotic Process Automation dominated the agenda. Today, AI goes significantly further. It is more efficient, more flexible, and far better suited to complex environments. Its strength lies in pattern recognition and in processing unstructured data, capabilities that traditional tools simply do not offer at scale.

As a result, AI is emerging as a lever for three distinct forms of value: enhanced strategic insight, greater operational efficiency, and more dynamic, real-time risk management, even as it introduces new categories of risk that CFOs must actively govern.

1. AI and the strategic value of looking ahead

The shift from a purely backward-looking function to predictive finance marks a fundamental transition. In a fast-changing environment, historical patterns that explain the past are no longer sufficient. Organizations need to look forward, continuously recalibrate expectations, and anticipate what comes next. Use cases such as rolling forecasts and cash flow optimization are taking on a new level of relevance and sophistication.

While many companies already run rolling forecasts, AI-driven models provide a far sharper lens. They combine internal historical data with external signals, including macroeconomic indicators. More advanced models can even incorporate variables such as weather patterns and other relevant data. The result is a more dynamic and accurate view that enables organizations to adapt continuously and make better-informed decisions.

2. A necessary push on efficiency

This is not an easy shift. Many organizations are struggling to execute, even as they recognize it as the only viable path forward. The pressure is most acute in the drive for structural efficiency, particularly across transactional activities and core processes such as accounting, accounts payable (AP), and accounts receivable (AR). These are precisely the areas where AI can deliver immediate and measurable impact.

Across the market, most companies start there. The use cases are clearer, easier to quantify, and supported by a well-defined return on investment.

Invoice processing is a case in point. Where OCR tools once dominated, AI-driven systems are rapidly taking over, capable not only of extracting unstructured data from PDFs, emails, and contracts, but also of interpreting that data and connecting it across sources.

This ability to understand context and synthesize multiple inputs represents a step change. AI excels at analyzing and processing unstructured data, and it identifies relationships across documents far more effectively than traditional tools ever could.

The implications for the transactional backbone of Finance are significant. Activities that were historically offshored to Shared Service Centers in lower-cost locations are now increasingly handled by AI agents. The cost base shifts materially. At the same time, the talent model evolves: demand moves toward profiles that are more business-savvy, analytically strong, and capable of working effectively with AI-enabled tools.

Concrete applications, tangible impact

New use cases are also emerging in credit management. By analyzing payment behavior and identifying patterns, companies can act far more proactively. Customers who consistently pay late can be targeted automatically, for example, with AI agents sending reminder emails ahead of due dates, without any manual intervention.

Similar applications are gaining traction in operational environments. At a logistics player in the fresh supply chain, AI agents are deployed to analyze and classify incoming complaints via email. What was once a labor-intensive process is now handled in seconds. The agent sorts and prioritizes comlplaints, resolves standard inquiries automatically, and routes exceptions or more complex cases to the appropriate staff.

Control remains critical

This evolution highlights a fundamental condition: trust in AI requires control. Systems must be rigorously tested and continuously monitored to prevent errors and so-called “hallucinations.” Only when quality is consistently assured can further automation be deployed with confidence.

It also underscores a broader point. The path to predictive finance is not just a technology transformation, it is equally an exercise in governance, process redesign, and risk management. That remains the common thread: how AI can deliver sustainable, measurable value within the Finance function.

3. Real-time Risk Management

AI enables a step change in how companies manage risk. It allows for real-time fraud detection, anomaly identification, and continuous control. Where controls today are often sample-based and retrospective, AI makes it possible to monitor transactions on an ongoing basis.

At scale, AI can review millions of entries and assess transactions in real time, flagging irregularities such as duplicate payments, unusual patterns, or errors in VAT reporting. Outliers in expense claims can be identified instantly. Detection becomes faster, more consistent, and far less dependent on manual intervention.

The real advantage lies in scale and speed. What was once manual and fragmented can now be executed continuously and comprehensively.

Audit processes also evolve. AI can support auditors by performing selected account checks, reducing audit effort and cost. At the same time, it strengthens the CFO’s position on compliance and financial integrity.

Taken together, these capabilities free up capacity within Finance. They allow the function to evolve into a true business partner, one that actively supports decision-making, looks ahead, and drives greater value creation across the organization.

2. Challenges for the CFO

AI is not Plug-and-Play

AI offers clear upside: efficiency gains, better decision-making, and new forms of value creation. But beyond the hype, a set of structural challenges quickly emerges. Four questions consistently come to the forefront.

• What is the ROI, and how do you measure it?
• Is your data and IT landscape ready to support it?
• How to manage risk, compliance, and ethical considerations?
• Does your organization have the right skills and culture to make it work?

1. The ROI Paradox

AI projects behave fundamentally differently from traditional software implementations such as ERP systems. Conventional SaaS solutions typically come with a clear upfront business case and a predictable cost model, often based on per-user licensing. AI does not.

Its cost structure is inherently variable. Most AI models operate on consumption (tokens, compute, and data volume). As usage scales, operating expenses can increase rapidly and unpredictably, driven by the number of interactions, the volume of data processed, and the computational intensity required to generate outputs. Budgeting, particularly in early stages, becomes less straightforward as usage patterns are still forming.

On the value side, the challenge is equally significant. Returns are rarely immediate or easily quantified in financial terms. Much of AI’s impact sits in less tangible benefits: time savings, improved decision quality, and enhanced customer experience. CFOs are therefore asked to approve business cases where the financial upside cannot be fully specified upfront.

For C-level executives, this requires a different lens on investment. Traditional ROI logic is often insufficient. AI demands a broader definition of value creation, one that explicitly incorporates strategic positioning and operational impact alongside financial return.

2. Data integrity

Also for AI, the “Garbage In, Garbage Out” reality still applies: AI is only as good as the data it runs on. That is not new, but in practice, it remains a critical constraint.

When financial and operational data is fragmented across legacy systems, AI has limited traction. Building AI on top of a weak data foundation rarely delivers sustainable value. At best, outputs are suboptimal and predictions unreliable. At worst, it introduces additional complexity and risk.

For many organizations, the first meaningful AI investment is therefore not the model itself, but the structuring of the underlying data and system landscape. In reality, this often proves more complex than anticipated. Data needs to be cleansed, standardized, and enriched before it becomes usable at scale.

A recent client case illustrates the point. Following multiple acquisitions, the company operated several billing engines that were not fully aligned. Critically, there was no consistent unique identifier across systems, making reliable AI deployment impossible.

Take a simple example: an opportunity created in Salesforce is converted into a contract, after which services are delivered and invoiced. Without a consistent identifier across that entire chain, end-to-end analysis or automation is not feasible. The result is manual reconciliation. Data must be cleaned, linked, and rebuilt before it can support AI use cases.

It is a labor-intensive and costly exercise. But it is also a prerequisite. Without it, AI cannot deliver meaningful or scalable impact.

3. Risk management, compliance and ethics

When AI is applied across diverse data types, a fundamental question must be asked, particularly in Europe: is this fully aligned with the EU AI Act?

In ensuring compliance, the CFO has a critical role to play. Especially when working with sensitive corporate data, processing must take place within a controlled and secure environment. Employees cannot simply input company data into public AI tools. Appropriate governance frameworks are required, supported by clear policies and, just as importantly, consistent enforcement.

This is not only a technology issue, but a control and accountability issue.

4. Culture and the skills gap

Every wave of new technology triggers concerns about job security. That is a natural response to transformation. Within Finance, this tension is most visible in the transactional space: invoice processing, customer follow-ups, and other repetitive activities that are already partially automatable today.

This shift inevitably forces Finance professionals to reposition themselves. AI enables the reduction of repetitive work and creates space for more analytical and business-oriented responsibilities. In that sense, Finance talent cannot afford to ignore the shift; those who do risk becoming increasingly irrelevant in a rapidly evolving function.

At C-suite level, the initial focus is often on efficiency gains. This is logical. Use cases in this domain are easier to implement and translate more readily into traditional ROI frameworks. However, this creates a structural risk: if AI remains confined to efficiency improvements, its strategic potential remains largely untapped.

Many organizations therefore start with informal AI communities to drive experimentation from the bottom up. While this dynamic can be valuable, it often runs into scalability constraints. Without integration into formal processes and governance structures, fragmentation quickly emerges.

Within Finance, the strongest starting point for AI adoption is often the controller community. Their familiarity with tools such as Microsoft Excel and Microsoft Power BI, combined with solid business understanding and digital maturity, positions them well to bridge the gap between technology and practice.

Ultimately, the extent to which organizations succeed in combining bottom-up momentum with top-down governance will determine whether AI remains a set of isolated experiments or evolves into a structural driver of value creation.