AI talk at finance dinners

At finance dinners this year, one of the easiest ways to get a headcount request through is to say the new hire will multiply output with artificial intelligence, not just add labor. Deloitte’s fourth-quarter 2025 Chief Financial Officer Signals survey found 87% of chief financial officers expect artificial intelligence to be very or extremely important to finance operations in 2026. (deloitte.com) That shift is showing up inside the back office, not just in flashy demos. Gartner said 58% of finance functions were already using artificial intelligence in 2024, up 21 percentage points from 2023, based on a June 2024 survey of 121 finance leaders. (gartner.com) The jobs getting attention are the ones finance teams hate because they are repetitive and expensive to get wrong. Recent accounting coverage points to real deployments in invoicing, accounts payable, forecasting, and other workflow-heavy tasks where one bad field or one missed exception can delay cash or distort reporting. (journalofaccountancy.com) That is why the conversation has moved past “buy an artificial intelligence tool” and toward “build one around our mess.” Finance teams often have custom commission plans, special billing rules, and approval chains that do not fit a generic software-as-a-service template. (journalofaccountancy.com) Commercial software is still part of the stack, but buyers want it to bend. Ernst & Young said finance leaders in its 2026 roundtables were focused on practical adoption, governance, and productivity, which is another way of saying they want artificial intelligence that can be configured, checked, and limited instead of a black box making unsupervised decisions. (ey.com) The caution is not theoretical. The Journal of Accountancy reported in February 2026 that finance leaders still validate outputs from more autonomous artificial intelligence systems because accuracy concerns remain, especially when the numbers feed forecasts, close processes, or compliance work. (journalofaccountancy.com) Pricing is also changing the buying math. As more vendors charge for artificial intelligence by usage instead of by seat, finance teams are being asked to manage software bills that behave more like utility bills, with costs rising when models are called more often. (pymnts.com) So the pitch winning rooms right now is not “replace the finance department.” It is “keep the finance system you trust, add controlled artificial intelligence to the painful steps, and give the vendor enough flexibility to match your commission logic, billing quirks, and approval rules.” (ey.com) That is why dinner-table artificial intelligence talk in finance sounds so specific. When buyers start talking about commissions, billing, close cycles, and exception handling instead of chatbots in the abstract, they are telling vendors exactly where the budget is moving. (gartner.com)