Agentic AI is quickly gaining attention among CFOs at large U.S. enterprises, with adoption and testing of this autonomous technology making significant strides in just three months. According to a recent survey by PYMNTS Intelligence, more than one in ten companies with revenues exceeding $1 billion began testing or using agentic AI by the start of the third quarter, marking a sharp shift in sentiment toward this emerging form of artificial intelligence.
The survey revealed that in May 2025, no U.S. enterprises had either deployed or piloted agentic AI. At that time, only 1.7% of CFOs were exploring its adoption. By July, however, the landscape had shifted: 6.7% of firms reported live usage, 5% had begun testing, and 8.3% were actively exploring adoption over the next 12 months. In total, 35% of CFOs surveyed indicated their companies were now using, piloting, or considering the technology.
This rapid momentum reflects a growing interest in agentic AI, which differs from generative AI by enabling systems to act autonomously within defined parameters. CFOs view this shift as an opportunity to streamline and automate tasks traditionally managed by teams of employees or requiring multiple layers of approval.
While enthusiasm for agentic AI is rising, skepticism still prevails. In May, 85% of CFOs surveyed expressed no plans to adopt agentic AI. By July, that figure had dropped to 65%, highlighting a growing but cautious interest in the technology. The reluctance to fully embrace agentic AI stems largely from concerns about granting it access to sensitive financial data and permissions for critical tasks.
For instance, less than half (45%) of CFOs were open to granting even limited access to sensitive directories, and only 8.3% were willing to provide moderate access. Among organizations already adopting agentic AI, these numbers improved slightly, with 43% open to moderate access and 57% to limited access. However, the trust gap remains a significant barrier to broader implementation.
CFOs are strategically testing agentic AI where it can provide the most value. Financial planning and analysis (FP&A) stands out as a top priority, with 70% of CFOs expressing strong interest in applying agentic AI to this function. Other areas of focus include financial reporting (68%), cost management (63%), and working capital optimization (63%). Strategic planning, decision support, and capital management also feature prominently in their plans.
Specific use cases are already emerging across industries. In education, agentic AI is being explored to automate board-level financial reports, while the healthcare sector is using it to reconcile accounts payable and receivable in real-time. Technology companies are applying it to benchmark spending against industry standards, and real estate firms are using it to create forecasts updated with market trends.
The survey also shed light on how organizations are approaching agentic AI adoption. Among firms testing or using the technology, 71% are building in-house capabilities to retain greater control and customization. Simultaneously, 43% are partnering with external vendors, fintechs, or consultants, while the same percentage are procuring off-the-shelf tools. Many companies are opting for a blended approach, experimenting with multiple models before committing fully.
In-house development offers long-term control but requires significant upfront investment and time, which may deter smaller or mid-sized companies. External partnerships, on the other hand, allow for faster deployment but carry risks such as overspending or acquiring tools that fail to deliver meaningful results.
Despite the growing interest, CFOs remain cautious about deploying agentic AI for high-stakes processes. For example, only 32% of respondents expressed strong interest in using agentic AI for treasury management, where errors could lead to financial and legal repercussions. Similarly, areas like tax compliance and risk management are approached with hesitancy due to the potential for unintended consequences.
Moreover, CFOs are grappling with the question of trust. For agentic AI to function effectively, it requires access to sensitive data and operational permissions - an area where many finance leaders remain hesitant. The survey underscores that until these trust issues are addressed, adoption will likely remain partial.
The surge in Q3 adoption is a notable milestone in agentic AI’s journey from concept to real-world application. While CFOs are showing increasing interest in testing and integrating the technology, their cautious approach to granting full autonomy reflects the complexity and stakes involved in financial operations. As more companies refine their strategies and early adopters report returns on investment, the trajectory of agentic AI’s adoption is poised to accelerate.
Ultimately, the future of agentic AI in corporate finance will hinge on how effectively CFOs navigate the delicate balance between opportunity and risk. This ongoing narrative will determine whether agentic AI evolves from pilot projects into a transformative tool for operational efficiency and decision-making.