The networking equipment company says it is focusing on four key areas, and launched a series of pilots to detect fraud with intelligent monitoring; provide intelligence forecasting; generate data insights; and make it easier to process complex documents.

But even amid all of that ongoing work, Scott Herren, Cisco EVP and chief financial officer, is still thinking ahead. “What are the next set of challenges that we don’t see today, that AI could be a huge help [with], too?” Herren asked rhetorically, as he joined other CFO leaders at a virtual panel discussion hosted by Fortune. “We’re still working on that.”

CFOs like Herren are confronting the changing relationship between humans and technology, said Joseph Fuller, a professor of management practice and cohead of the Managing the Future of Work project at Harvard Business School. For decades, Fuller explains, businesses sought to give data to leaders to empower them to make better decisions.

But AI-based tools are emerging to sort through multiple sets of data and then present those recommendations to people. The responsibility of the worker is shifting to approving the decisions that are being made by machines. This shift has inverted the relationship between humans and technology.

For CFOs and finance departments, it means that rather than performing hindsight analysis on past trends—including pricing, sales, and market share—and trying to make predictions, AI tools are being used to give CFOs an earlier warning so they can better plan for risks.

“You’re going to have the opportunity to detect problems much earlier and therefore manage your risk much more effectively,” said Fuller. “Whether that’s a supply-chain risk, or overstock risk, or losing market share because of pricing, all of those things will become much more visible to the CFO organization so they’re not limited to retroactive, late interventions to correct inadequacies in performance.”

Joy Mbanugo, CFO at IT service management company ServiceRocket, said she has been encouraging her team to embrace AI. And in some cases, younger employees who are especially eager to use AI to enhance their work were even ahead of her own thinking on the technology. 

“For some people who are embracing AI, yes, they’re becoming more strategic,” said Mbanugo.

She sees great potential in AI helping in practical ways, like the use of a chat feature that can convey complex topics in multiple languages for ServiceRocket’s team members based in various markets around the globe. AI can be used to draft internal policy changes. AI is also being used to better predict revenue.

“I think, ‘What’s cutting edge? And what will help finance teams … make better decisions … being able to take all that data and very quickly—not taking days, but taking minutes,’” said Mbanugo.

Checkr, which provides background checks for businesses, has been using AI since its inception a decade ago. But the launch of ChatGPT and other AI tools have lowered the barrier to use emerging technologies like generative AI, and can also lower the cost to access AI. 

For venture-backed businesses like Checkr, that’s important. Expectations have changed for startups, especially after interest rates were hiked as the U.S. government aimed to tackle inflation. That has put pressure on startups to be more mindful of their spending.

“What we’re finding is the intensity and pressure on demonstrating profitability is higher than ever,” said Naeem Ishaq, CFO at Checkr. “And so utilization of AI can dramatically improve profitability by automating tasks around things like first-line operations, helping to make your employees more productive, and ultimately contributing to the bottom line.”

Fuller said AI is so transformational that companies should design entire processes to maximize and exploit the gains that can be captured from the technology, and then organize their team members around the AI, as opposed to adding AI onto existing ways of work. 

He said the biggest risk that companies face today as they approach this AI-driven transformation is that they are proceeding too cautiously. Studies consistently align with his thinking. A recent survey by consulting giant EY found only 26% of chief information officers had deployed AI in any meaningful way and that less than half had even got to the point of a test project or proof of concept.

“You are incurring a big risk if you’re moving slowly and your archrivals are moving fast, particularly if they are larger,” said Fuller.