AI in the CFO Function: The Leadership Challenge No One Should Delegate

Artificial intelligence has moved beyond experimentation in finance. CFOs are no longer debating whether AI will matter. Instead, they are asking where it already creates value, how quickly it should be scaled, and what it means for the future of finance leadership and talent. Across our recent conversations with finance leaders, five observations stood out:

The most interesting use cases are no longer peripheral experiments. They increasingly touch the core of the CFO agenda: transactional efficiency, reporting, forecasting, compliance, margin improvement, and drivers of growth. Across the broader finance function, applications range from highly operational use cases to more advanced decision-support capabilities:

• Automating billing, invoicing, payment and accounting processes
• Generating rolling (monthly) forecasts and more dynamic performance outlooks
• Monitoring contract compliance 
• Identifying cost-reduction opportunities in procurement
• Supporting revenue growth by enabling sales and services teams

Two tangible examples for the business impact come from supplier management and purchasing: One CFO described how AI can make previously impractical contract reviews feasible by analyzing large volumes of agreements more quickly and systematically. Instead of relying on selective manual reviews, finance teams can identify deviations, reduce the risk of human error, and uncover meaningful savings opportunities in areas that were historically difficult to address at scale. In specific cases, small investments in AI support led to multi-million refunds due to incorrectly issued invoices.

The same logic applies to purchasing. AI can help compare supplier pricing across geographies, business units, and historic negotiation cycles. Procurement teams can use these insights to identify patterns, benchmark terms more consistently, and support renegotiations with stronger analytical evidence. In parts, AI initiates discussions with long tail suppliers that were not actively managed before.

Beyond finance, CFOs are also beginning to see AI as a lever for revenue growth. AI can help employees handle more complex customer conversations, improve responsiveness, and compensate for talent bottlenecks. Rather than making employees obsolete, AI can expand their capabilities and increase their commercial relevance.

At the same time, CFOs are beginning to manage the spend on AI itself. Some CFOs are actively managing token costs comparing different AI providers for every usage and establishing a proper approval and review process for AI projects. In addition, they provide each employee with a specific monthly token budget. This creates transparency around AI consumption while still encouraging usage within a controlled enterprise environment.

Public discussions about AI often focus on efficiency through automation. Many CFOs emphasize something broader: AI enables finance teams to evaluate more scenarios and test more assumptions than previously possible. Rolling forecasts can be updated continuously or ad hoc rather than periodically. Alternative scenarios can be simulated within minutes rather than days.

As a result, organizations are simultaneously achieving:

• More comprehensive scenario planning
• Higher-quality decision support
• Faster insights
• Lower manual effort

For many finance leaders, this is the path they always wanted to take: the finance function is evolving from a reporting and planning engine into a real-time decision-support capability.

Every AI discussion ultimately comes back to data. The consensus among CFOs is clear: large-scale AI adoption requires not only access to data, but also a common, integrated, and well-governed data foundation.

The issue is therefore twofold. Organizations need to make the right data accessible while also ensuring that it is consistent, clean, and logically organized. AI cannot create clarity where definitions differ, master data is fragmented, or unresolved data artefacts continue to circulate through the system.

As one CFO summarized: "If different parts of the business use different definitions of KPIs, AI simply scales the confusion."

Harmonized KPI definitions, common master data, integrated systems, and clear ownership therefore remain critical prerequisites.

At the same time, finance leaders object to turning data quality into an excuse for inaction. Therefore, many organizations are launching targeted use cases while continuing their broader data-transformation journey. They are learning through experimentation, improving data discipline along the way, and building momentum long before the full data landscape is perfect. And in addition, they use AI to speed up the clean-up and data harmonization to build the ground for the usage of AI on a broader base.

When it comes to implementation, many CFOs describe a similar pattern: AI tools are made broadly available, yet adoption varies significantly between teams. The differentiator is not the software alone. It is leadership. The real challenge is getting the organization to embrace it. 

Organizations that achieve strong adoption tend to follow a similar playbook:

• Make AI broadly available.
• Position AI as an opportunity, not a threat.
• Encourage experimentation.
• Create internal challenges and learning formats.
• Share success stories openly.
• Reward curiosity and knowledge sharing.
• Establish a system to monitor usage and manage AI projects.

In many finance functions concerns from employees and employee representatives decrease significantly when AI is introduced as an offer rather than as a top-down efficiency initiative. This positioning matters: if companies do not provide secure, enterprise-wide tools, employees may still use AI privately, creating the risk that sensitive company data is entered into personal accounts. 

Another adoption-related tension is model availability. Many finance teams initially were frustrated that they did not always have access to the latest AI model. Over time, however, the organizations learned that for every day finance use, and likely for the vast majority of use cases across the CFO function and the broader organization, the model already available was often good enough. The bigger challenge was not always having the most advanced model but helping people use the available tools confidently and consistently.

People tend to become advocates once they experience the practical benefits themselves. The leadership challenge is therefore not selecting the best AI solution; it is creating an environment in which people feel safe to experiment, learn, and adopt new ways of working.

As one CFO put it: "The bottleneck is no longer computing power. It's organizational willingness to change."

As a consequence, we see an increasing gap in the performance of finance organizations that is quickly growing between early AI adaptors and the ones struggling with the first steps.
 

The final observation concerns talent. When asked what CFOs increasingly look for in future finance leaders, one characteristic surfaced is curiosity.

Technical finance expertise remains essential. Yet the willingness to explore new technologies, challenge established ways of working, and learning continuously is becoming just as important.

This has implications not only for selection but also for development.

Several CFOs highlighted that junior professionals will not necessarily need to perform every process in the future. The goal is not to calculate everything by hand. "Nobody calculates regressions manually anymore. But everyone needs to understand when the result doesn't make sense."

The same applies to AI. Finance professionals must understand the logic behind outputs, recognize limitations, challenge assumptions, and know when further investigation is required. The skill is shifting from execution to interpretation. Future CFOs will not differentiate themselves by producing information faster; they will differentiate themselves by asking better questions, exercising stronger judgment, and making better decisions.

This means, while future CFOs still need to learn the different subfunctions, they might not need to stay in each function for several years to have the basis for good judgement. This may open opportunities to develop CFO candidates even more broadly across the entire value chain, not just in the financial core functions. 

When AI can improve the productivity of the basic transactional finance processes, identifies opportunities for growth and increased profitability, the new defining question for finance leaders becomes different:

How do we build organizations, leadership teams, and future CFO talent that can fully capture this opportunity?

That, more than the technology itself, may be the real AI challenge for today's CFOs.