Are Finance Teams Shrinking?

AI is changing the way accountants allocate their time. A study conducted by Stanford and MIT found that accountants using generative AI supported 55% more clients per week, shifted about 8.5% of their time away from routine data entry toward higher-value tasks and closed their monthly books 7.5 days faster.

Kopp argues that the effect won't necessarily be fewer accounting jobs overall, drawing a comparison to software engineering: “There are more engineering jobs that are required today thanks to AI, not less. I think the same will happen in accounting.”

Finance executives are already seeing a shift in how much work smaller teams can handle. In KPMG’s Adaptability Survey, 62% of CFO and chief tax officer respondents cited quantifiable cost savings and efficiencies from what the firm described as “smaller teams with bigger outputs.”

AI-hiring startup Mercor offers a more extreme example. The company is using Rillet while operating a business with more than $2 billion in annual recurring revenue with a finance team of only three people. Kopp said he isn't seeing companies install Rillet and fire their accounting teams. “We don't see our customers firing their teams, rather being able to be selective in hiring and save costs that way,” he said.

There is a tension between accountants resistant to adopting these tools and companies trying to grow without adding the same amount of overhead. Matt Twiford, a fractional CFO and managing director of Pegacorn Group, said some lower-level accounting work is already approaching full automation. “The AP data entry and coding is at a point where it can almost be fully automated,” Twiford told me in written comments. “As a result, we are seeing the need for lower-level roles being reduced, and it’s working its way up the career ladder as well.”

Jamie Corby, CEO of fractional finance and strategic accounting firm Corby & Associates, sees the effect differently. His firm has continued adding staff after putting agents into production. “It's just re-stacking the work,” Corby said in written comments when asked whether AI could replace human accountants. “We've continued to add staff since putting agents into production. However, what has changed is what staff now do all day.”

His firm uses agents for daily bank reconciliations, transaction categorization and month-end close preparation, with the bigger change being leverage: the firm can now carry a client load that previously would have required a mid-sized back office.

Who Audits The AI Accountant?

As AI agents move directly into the general ledger and take on work that affects financial statements, auditability and governance become central to whether companies can use them at scale.

“There's a true opportunity here in accounting, especially when it comes against the governance pieces, making sure the audit logs are clean, the Big Four are bought into this, and there's a safe and well-auditable environment as part of the AI work that's being done,” Kopp said.

On the audit side, Rillet works with two of the Big Four, EY and KPMG, as well as more than half of the top 20 accounting firms in the U.S. Its alliance with EY focuses specifically on AI-native finance transformation with risk management and controls built into the process. The PCAOB, the regulator overseeing audits of U.S. public companies, is one of the bodies that will eventually have to adapt to how AI changes accounting and audit work.

To Kopp, accuracy, speed and auditability are the three things that matter most when applying AI to accounting. Rillet keeps deterministic accounting data alongside its AI systems and maintains detailed logs of the steps agents take — without them, customers and auditors cannot trust the system.

“Auditors need to be able to trace material transactions back to the system of record and reconstruct what the agent did, what information it relied on and why it reached a particular result,” Richard Chambers, senior advisor for risk and audit at Optro and former chair of UNICEF’s Audit Advisory Committee, told me in a written response. “AI can sound extremely confident even when it is wrong, so confidence should never be confused with competence.”

What’s Left For Humans?

Yousuf Rizvi, CPA and principal at Ridgeway Financial Services, believes AI does not remove the accountability structure that already exists in accounting. An agent can take the place of a preparer for some tasks, but it has no professional accountability of its own. “The agent can prepare. A person still has to be the one who owns it,” Rizvi told me in written comments.

As AI takes over more of the mechanics of accounting, the remaining human role is shifting toward reviewing the work, handling judgment calls and exceptions, and taking responsibility for the numbers. “Every single transaction, one way or another, has to be signed off by a human,” Kopp said. He sees the accounting workflow moving toward agents performing more of the underlying work and humans reviewing and approving the results.

Twiford sees a hard boundary around responsibility as well. “For publicly traded companies, the CEO & CFO personally attest that the financials are accurate and that internal controls are effective,” he said. “No matter how much of the work an AI agent does, that personal liability can't be handed off. A human executive is still the one signing their name to it and has to stand behind it.”

The shift, then, is in how the job is structured: AI saves hours on routine tasks, while accountants move from doing every step of the work to reviewing, directing and taking responsibility for work performed by AI.