From record-keeper to architect of value: the skills that will matter most in the era of agentic AI

For decades, a company’s finance function has been synonymous with rigor, control, and a backward-looking gaze: month-end closings, reconciliations, reports that confirm what has already happened. The chief financial officer was, in essence, the guardian of the numbers. Today, that picture is changing fast. Artificial intelligence is not coming to eliminate the financial role, but to reshape it, from reporting past performances, toward anticipating and shaping future outcomes.

It is a transformation that carries a peculiar paradox: the CFO is being asked to fund the very technology that is redefining their own role. For that reason, how finance leaders understand and orchestrate AI over the coming years will make the difference between companies that gain a durable competitive advantage and those that remain stuck in fragmented automation with no real impact.

A turning point, not a passing fad

The numbers confirm that we are no longer talking about isolated experiments. According to a Gartner survey, the share of finance leaders using artificial intelligence within the financial role rose from 37% in 2023 to roughly 59% in 2025. McKinsey, in turn, reports that 44% of said CFOs in 2025 were using generative AI for more than five use cases, a dramatic jump from just 7% a year earlier.

And yet, organizations should avoid premature enthusiasm and carefully assess its limitations and implementation challenges. Nearly two-thirds of organizations admit they have not yet begun scaling AI across the enterprise, and most report only low-to-moderate impact in the early stages. The message is clear: the technology is available and accessible, but value does not appear automatically. It appears where AI is tied to concrete business needs, not adopted for the sake of innovation.

From generative AI to agentic AI

The first phase of the current wave was dominated by generative AI, assistants that summarize documents, generate analytical reports, and answer questions based on internal data. The next phase, already on its way, is the implementation of agentic AI: systems that not only suggest, but autonomously carry out complex tasks, from monitoring invoices to proposing budget adjustments, under human oversight.

The difference is fundamental. In the era of generative AI, the challenge was validating the calculations. In the era of agentic AI, the challenge becomes validating the assumptions: defining decision boundaries, training the guardrails, and stress-testing how the agents behave in edge cases. It is no coincidence that more specialists speak of a “human-in-the-loop” era of finance, in which advanced automation and human judgment work together, not one in place of the other.

What artificial intelligence automates in finance

To grasp the stakes, it helps to look at concrete applications with measurable impact. McKinsey research documents several such cases, reported, as is standard in this kind of study, without naming the companies involved, that show AI is not an abstract promise but a tool with a tangible effect on both money and time:

• Planning and analysis (FP&A). A global consumer goods company cut the time spent on budget variance analysis by roughly 30% using a generative AI assistant.
• Working capital management. An agentic system that automatically matches invoices against contracts identified, at a biotech company, compliance issues equivalent to about 4% of spend a potential recovery of tens of millions of dollars a year through the correct enforcement of discounts and rebates.
• Cost optimization. A European financial institution reduced costs by roughly 10% of its spend base by automatically categorizing invoices and identifying inefficiencies in energy, travel, and facilities management.
• Knowledge management, payments, and anomaly detection. At the level of current usage, the most common applications remain organizing information (around 49%), accounts-payable automation (37%), and error and discrepancies detection (34%).

What connects these examples is a common principle: AI takes over the repetitive, high-volume tasks governed by clear rules, freeing up time for the work of analysis, interpretation, and decision-making, precisely the area where a finance professional’s value is hardest to replace.

The CFO role: from record-keeper to architect of value

As routine tasks become automated, the center of gravity of the CFO role shifts as well. The shorthand for this transition, “from record-keeper to value creator” — captures the new reality well. The CFO is no longer merely the one who reports and controls but becomes the enterprise-wide sponsor of intelligent automation and an architect of how the company creates value.

This shift also brings new expectations to the board’s levels. Promises such as “we saved time” are no longer enough. Boards now demand a specific impact on the profit-and-loss statement and new, outcome-oriented metrics: the decision-making agility, the errors reduced rate, the organization’s ability to adapt. The focus is increasingly shifting from the effort invested to the value and outcomes delivered.

The skills that will matter most

If technology takes over technical and repetitive work, then the difference between professionals will be made in the areas of skill that AI cannot replicate. Four categories emerge as essential:

1. Critical thinking and data interpretation

In a world where anyone can generate a report in seconds, the advantage no longer lies in producing the numbers, but in asking the right question and interpreting them. The financial professional becomes the one who validates the assumptions, spots what is missing from an analysis, and translates algorithmic outputs into business decisions.

2. Technological fluency and “data fluency”

Not every specialist in finance needs to become a programmer, but understanding how the models work, what their limits are, and how to frame a question is becoming a core competency. Indeed, the barrier most often cited in AI adoption remains precisely the gap in skills and data literacy.

3. Communication and influence

The role of strategic partner requires the ability to explain, to persuade, and to build bridges between finance and the company’s other functions. A valuable insight that is not communicated persuasively remains without effect.

4. Ethics, judgment, and accountability

As decisions are increasingly supported — or even executed — by algorithms, human oversight remains essential to ensure the quality, integrity and accountability of decision-making. Ethical judgment cannot be automated; on the contrary, it becomes all the more valuable.

Data as a strategic asset

No model, regardless of its sophistication, can compensate for poor-quality data. This is perhaps the least spectacular but most important lesson of the current wave. The CFO is called upon to become a genuine “data steward” to break down the information silos between departments and to build a single, coherent, trustworthy source. Without that foundation, investments in AI generate, at best, inaccurate results.

Governance and ethics: from an obligation to a competitive advantage

For a long time, governance and compliance were seen as a necessary cost, a defensive brake. In the era of agentic AI, they are turning into a strategic differentiator. A transparent, controllable, and ethical deployment of artificial intelligence becomes an argument in front of investors, regulators, and clients. Companies that can demonstrate their control upon algorithms – rather than the other way around -will inspire greater trust and gain easier access to capital and partnerships.

Pitfalls to avoid

The experience of the organizations that have advanced to the furthest points, are teaching us a few recurring mistakes that finance leaders can anticipate:

• Waiting for the “perfect” data before implementation – a sure recipe for paralysis.
• Attempting total transformation in one single move – instead of well determined steps with demonstrable impact.
• Pilots without a roadmap, disconnected from the real priorities of the business.
• Underestimating change management – real implementation depends on people, not just on technology.
• Automating fragmented processes before standardizing them – that only automates chaos at greater scale.

Conclusion: technology amplifies, but does not replace, the human factor

The future of corporate finance is not a story about robots replacing people, but about professionals who, freed from routine, climb the value ladder. The CFO of the upcoming decade will not be defined by the generated reports, but by strategic clarity provided; not by how many numbers they master, but by how many good decisions they help the company make.

Paradoxically, the more powerful technology becomes, the more important the deeply human qualities grow; critical thinking, integrity, the ability to communicate and to inspire trust. The companies that understand that AI is a tool for amplifying human judgment, and not a substitute for it, are the ones that will write in the next chapter of corporate finance.