Most generative AI projects in finance fail. Here’s what it means for CFOs and their future hires.
Mid-market finance chiefs aren’t hiring like they used to. A Gartner survey of more than 300 finance executives shows that while companies are pouring more money into artificial intelligence tools, the share of CFOs planning to expand headcount has fallen.
“Four out of five were either freezing capacity or headcount or reducing capacity in their team,” Alok Ajmera, chief executive officer at Prophix, told Global Finance in a phone interview.
While CEOs and boards pressure their teams to show productivity gains from generative AI, results have been uneven, Ajmera added.
The GenAI Reality Check
Another Gartner estimate shows that more than 90% of generative AI proof-of-concept projects in finance departments failed to generate incremental value. This “staggering amount,” Ajmera said, curdles AI enthusiasm into AI skepticism.
“A lot of projects from an AI perspective felt really good on paper, but have not actually materialized the value … in real life,” he added.
He attributed the shortfall to a mismatch between the technology and finance work itself.
“This is not a probabilistic exercise, this is a deterministic exercise,” Ajmera said. “You can’t be 99% accurate with your numbers. You have to be 100% accurate.” CFOs remain comfortable using AI for reporting, commentary and analytics, he said, but “extraordinarily uncomfortable” letting it touch journal entries or adjust numbers directly.
Shifting Skills, not Mass Layoffs
Ajmera pushed back on warnings of mass AI-driven unemployment, including recent comments from Amazon founder Jeff Bezos, saying “the Doomerism view has been overhyped.” He pointed to software engineering — home to agentic coding, the most monetized AI use case to date — as evidence. “We have hired more engineers in 2026 than we did in 2025,” despite productivity tools making individual engineers more efficient, he said.
He predicted finance will see similar skill displacement rather than outright job losses. “I would not be surprised in a couple of years if we start seeing finance operations engineers” managing AI agents on staff. Slower hiring, Ajmera added, is not the same as letting people go.
Ajmera also described a broader consolidation trend, as companies unwind software sprawl built up earlier this decade. He cited a mid-market manufacturer in the Midwest whose cloud application count grew “from five or six applications to 25 or 30” before Prophix helped consolidate roughly nine or 10 of those tools onto a single platform.
Looking ahead, Ajmera said he expects more caution from CFOs. “There’s a lot of caution in the air,” he said, predicting longer purchasing cycles and heavier scrutiny of technology spending amid broader economic uncertainty.
Anthony Noto covers corporate finance and private credit. Contact him at anoto@gfmag.com