New Report: Over 10% of CFOs Now Testing Agentic AI

If you had asked a Chief Financial Officer in March of this year about “agentic AI,” you probably would have been met with a blank stare. The technology—AI that can autonomously plan, reason, and take multi-step actions—was largely a conceptual buzzword.

Fast forward 90 days.

According to a report from PYMNTS.com, by this past July, over 11% of enterprise CFOs in the U.S. were already using or actively testing agentic AI.

In the span of a single quarter, a revolutionary technology went from zero to live test runs in one of the most risk-averse, security-conscious, and critical departments in any business: Finance.

What Is Agentic AI (And Why Do CFOs Care)?

We’ve all gotten used to generative AI, like ChatGPT or Gemini. You give it a prompt, and it generates a response. You are still in full control.

Agentic AI is a different beast. You don’t just give it a prompt; you give it a goal.

• Generative AI: “Write me an email summarizing our Q3 sales data.”

• Agentic AI: “Analyze our Q3 sales data, compare it to the marketing budget, identify our top three most profitable customer segments, and draft a presentation for the board meeting.”

The agentic model can plan that multi-step task, execute the analysis, access different data sources, and produce the final output, all without human intervention at each step.

It’s no wonder CFOs are intrigued. The PYMNTS.com article, “How Agentic AI Went From Zero to CFO Test Runs in 90 Days,” notes that executives are most interested in using it for complex tasks like:

• Financial planning and analysis
• Financial reporting
• Cost management

Imagine an AI agent that can reconcile thousands of transactions, flag anomalies, and prepare financial reports in real-time. That’s the promise that has captured the C-suite’s attention.

The Trust Barrier

While the speed of adoption is breathtaking, the report highlights a critical and perfectly logical wall that adoption has hit: trust.

While CFOs are eager to let agents analyze and report, they are nowhere near ready to let them act. The study found that executives are deeply cautious about using agentic AI for high-stakes functions like:

• Treasury management
• Risk and compliance
• Autonomously initiating payments

This reveals the “agentic trust gap.” No CFO in the survey was willing to grant an AI “substantial or full access” to sensitive internal data like vendor lists, and even those already piloting the tech were hesitant.

This is the great paradox: for an AI agent to be truly useful in finance, it needs access to the company’s most sensitive data and permission to act on it. But those are the very two things a CFO is paid to protect.

The Takeaway for Your Business

This “zero to CFO” phenomenon is a case study in our new reality. The adoption cycle for powerful AI has collapsed from years to months.

Companies that have already seen a positive return on investment from generative AI are, unsurprisingly, the first to jump on agentic AI. Success is breeding confidence, and confidence is breeding speed.

It shows that the “wait and see” approach to AI is becoming a high-risk strategy in itself. If your competitors can successfully automate complex analysis in a single quarter, what will they be able to do in the next?

The trust barrier is real and necessary. However, the question now is no longer if these tools will reshape business, but how to build the guardrails of trust quickly enough to keep pace.

Navigating the AI Frontier

The “trust gap” is the single biggest challenge for leaders evaluating autonomous AI.

You don’t have to choose between innovation and safety. We specialize in helping finance and insurance teams build the robust guardrails, custom permissions, and secure data-handling policies needed to pilot and scale AI responsibly.

Schedule a consultation with us to build your trusted corporate AI framework.