A new report on agentic AI in finance, tax, and accounting finds that while the majority of teams are planning to invest in automation most aren’t delivering.
The study from automation platform Savant Labs, based on responses from finance leaders across 22 industries in North America, finds that while 75 percent of respondents overwhelmingly cite 2026 as the year for strategic agentic automation and investment, a surprising 63 percent have either no adoption or basic implementation in one to two functions, suggesting their aspirations are lagging AI execution.
The research identifies a pronounced governance gap as the most significant barrier to execution, with 36 percent of leaders reporting persistent concerns around the ability to govern, audit and control AI automation workflows, particularly when those workflows interact with core ERP systems and sensitive financial data.
“Enterprise teams are aligned on the expected benefits of agentic automation, but many companies are struggling to operationalize it,” says Chitrang Shah, CEO and co-founder of Savant Labs. “Their uncertainty around governance, auditability, and risk is holding back confident scaling — which can be addressed with the advancements of agentic automation platforms like Savant.”
Only six percent of leaders say they have an advanced strategy with broadscale implementation across their teams, with the majority of respondents admitting their organizations’ adoption remains limited to isolated pilots or individual single function use cases.
Governance and trust are seen as the primary blockers. Over 60 percent of the leaders rank data governance and security, along with ERP integration concerns, as the primary blockers to widespread adoption. This concern ranks five times higher than ‘cost and ROI’ and three times higher than concerns over quality control and lack of team skills combined.
There’s no anticipation of increased staffing levels, 82 percent of respondents expect no net headcount change from AI in 2026. Rather the shift is in skill expectations suggesting increased pressure on existing teams to deliver more output with the same resources.
The full report is available from the Savant Labs site.
Image credit: Alexandersikov/Dreamstime.com
