Agentic AI Gains Strength and Redefines the CFO Role

Agentic AI has moved beyond being an experimental concept to becoming an emerging priority in corporate finance. According to the Agentic AI Jumps From Hype to Reality report (PYMNTS, 2025), CFOs’ interest has accelerated significantly in just two months: in May, only 1.7% considered adoption, and 85% had no plans.

By July, the picture changed drastically:

Altogether, more than one-third of CFOs surveyed are now at some level of openness toward Agentic AI. This shift shows that the debate has moved from theory to practice, positioning intelligent agents as a strategic tool in the financial leadership toolkit.

The breakdown by financial functions reveals that CFOs prioritize Agentic AI in areas where it adds intelligence and reduces analytical friction, while being more cautious in those involving the direct movement of money.

The breakdown by financial functions reveals that CFOs prioritize Agentic AI in areas where it adds intelligence and reduces analytical friction, while being more cautious in those involving the direct movement of money.

High-Interest Functions (Over 60% “Very Interested”)

Here, CFOs see Agentic AI as a trusted copilot. These are data-intensive tasks where errors don’t immediately lead to cash losses. In other words: the ideal lab to test agents without taking excessive risks.

• Financial planning and analysis (70%)
• Financial reporting (68.3%)
• Cost management (63.3%)
• Working capital optimization (63.3%)
• Strategic planning and decision support (63.3%)

The logic is pretty simple: low operational risk + high information volume = perfect ground for agents to demonstrate real value.

Mid-Interest Functions (45–50% “Very Interested”)

In capital, risk, taxation, and corporate governance, CFOs recognize the potential but know that here, a mistake doesn’t just cost money—it can mean regulatory penalties or damage credibility with investors and auditors.

• Capital management (50%)
• Tax planning and compliance (45%)
• Risk management (45%)
• Corporate governance (46.7%)

Interest is present but moves cautiously: high relevance + high regulatory risk = slower, more fragmented adoption.

High-Trust Functions (Under 35% “Very Interested”)

Treasury is the area where CFOs remain the most cautious. Although 53.3% show partial interest, only 31.7% declare themselves “very interested.” Here, the barrier isn’t technology but trust: granting access to payments requires guarantees of control, traceability, and reversibility.

For many CFOs, treasury isn’t a “no,” but the last frontier of trust. Agentic AI will be able to cross it once it proves it can operate under the same security standards that currently underpin the most critical financial systems.

The conclusion is undeniable: Agentic AI begins its journey in analytical intelligence, but the next cycle will be transactional. That step will demand full confidence in security, traceability, and governance controls at the level corporate finance requires.

This leaves a dual takeaway for the industry: adoption is no longer nascent — nearly one-third of CFOs are using, testing, or planning to adopt it — and immediate value is concentrated in analytics. The challenge now is to support this initial cycle, build trust, and prepare the ground so that automation also reaches treasury and payments.

At Prometeo, we share this vision and are working on the banking infrastructure that will enable this transition toward AI agents capable of operating payments, validations, and treasury in real time.