Finance leaders are already putting AI agents to work.
They expect those agents to read, instruct and transact across banking workflows. How should institutions recognize, authorize and govern agent-initiated activity?

Nearly 3 out of 4 finance leaders we surveyed say AI agents are important to their operations within the year, and if their bank won't support them, they will find workarounds. The complete report provides in-depth details:
Already use or are piloting AI agents in finance and banking work.
Expect AI agents to act in transact-level banking within 12 months.
Leaders will find workarounds if their bank can't support agent use.
Finance leaders currently use AI agents to some degree to automate banking work, and they have clear plans to expand use in the next 12 months.
Most banking systems are designed for human customers, not requests from their AI agents. This report is about that gap and the steps needed to close it.
Learn where agents are coming from and what they are doing today.
Check out the ranking of bank workflows finance leaders plan to involve agents in.
Learn if there is an automony gap between basic banking tasks and consequential actions.
Discover what's most important on their list.
Get all of the data inside the full report.
OvationCXM surveyed 520 U.S. finance leaders to learn how they use AI agents today, where they expect to expand use in banking, the authority they will delegate and what controls increase confidence in agent use. Respondents represented finance leadership roles: CFOs, Heads/VPs of Finance, Treasurers, Controllers and Finance/Accounting Managers, in companies with annual revenue ranging from < $5 million to $1 billion+.An AI agent was defined as an AI system capable of initiating and completing multi-step financial tasks within defined rules and approvals, rather than one that only provides information or generates content. Results are based on the respondents eligible for each question, so sample sizes vary. Multi-select responses may total more than 100%; percentages are rounded.For analysis, OvationCXM grouped selected banking activities into Read, Instruct and Transact categories, and conducted respondent-level cross-tabulations where noted. These are analytical groupings, not survey response options. Relationships between variables represent associations within the survey sample and should not be interpreted as causal.Centiment conducted fieldwork in August 2026.AI-agent usage is self-reported. The research does not independently verify the technical architecture or autonomy of the tools respondents identified. All product and company names are trademarks or registered trademarks of their respective owners. Their inclusion in this report is for identification and research reporting purposes only and does not imply endorsement or affiliation.