Are finance leaders moving too fast on agentic AI?

More than 90% of senior finance leaders are facing moderate to significant career pressure to demonstrate that agentic AI is delivering ROI, with the majority of respondents to a global survey painting a picture of speed often being valued over governance and accountability.

For the report Agents of Change: How the Race to Deploy AI Agents is Outrunning Financial Governance, Avalara surveyed more than 1,500 CFOs and senior finance leaders with recent agentic AI experience in four countries: the U.S., UK, India, and Australia.

Half reported significant pressure (and 42% reported moderate pressure) to demonstrate the ROI of agentic AI in the finance function and to do so fast. Twenty-nine percent described their focus as entirely on speed, and another 41% said it was mostly on speed. Just 7% said governance of AI was being prioritized over speed.

While nearly 90% reported some ROI, just 38% described it as “at scale” and 50% described it as “limited.”

“Finance leaders are right to move quickly to capitalize on agentic AI opportunities, but speed without accountability creates new forms of risk, and speed without rethinking workflows limits ROI,” Avalara CEO Hugo Sarrazin said in a news release. “The organizations that realize the greatest value from AI won’t simply deploy more agents. They’ll leverage agents with trusted data, governed workflows, and clear controls that enable automation with confidence.”

In the push for speed:

• 36% said they have no one specifically responsible for understanding how AI agents work, although more than half of those respondents were actively working to hire or develop staff.
• 23% said accountability for a significant AI error would be unclear or that such accountability sits with no single individual or team.
• 53% are very confident and 44% somewhat confident that they could give a regulator or auditor a clear, complete, auditable explanation of an AI agent’s actions.

A Grant Thornton survey earlier this year identified governance or compliance barriers as the leading cause of AI underperformance, and an accompanying report recommended steps leaders can take to close the so-called “AI proof gap.”

While the Avalara survey similarly signaled concerns about an overemphasis on speed in agentic AI adoption, it also found that most companies are maintaining people as a crucial check for agentic AI’s output.

In six of seven business processes profiled in the survey, less than 15% of leaders were having AI act autonomously, with humans brought in only to review exceptions. AI agents have advanced a bit further in financial planning and analysis, where 19% of AI functions are working autonomously. In all areas, the most common approach featured AI offering recommendations that humans must approve.

— To comment on this article or to suggest an idea for another article, contact Bryan Strickland at Bryan.Strickland@aicpa-cima.com.