How one company got its finance team to actually use AI

Sponsored by

Shake the (cap) table. Pulley is cap table management that works for you. They’re changing the game with ready-made reports and built-in compliance that can accelerate daily work.

Integrating AI into an organization is easier said than done—this is a lesson Cameron Kinloch, former CFO of enterprise AI developer Weights & Biases, learned firsthand. Despite leading a finance team at an AI development company, she still struggled to get her people to consistently use AI in their workflows.

As a CFO, Kinloch wasn’t alone; according to an August survey by OneStream of 350+ full-time CFOs in the US, UK and Australia, 75% said they’re in charge of their company’s AI strategy. And nearly 70% of companies have already added AI tools to their toolbox, according to a May survey from OwlLabs and Pulse. The pressure to see tangible success may continue to intensify, as 83% of CFOs OneStream surveyed expected an increase in AI investment at their organizations next year.

When it came to her team, Kinloch saw the warning signs. She would see them spending hours on mundane tasks, but when she asked if they’d consulted their enterprise ChatGPT, “the answer was always no.”

“I realized, no one’s actually taking the time to embed it in their workflows,” Kinloch, who left Weights & Biases in June 2025 and is currently the founder of True Vantage Partners, an advisory and fractional CFO firm, told CFO Brew. “I could see that people weren’t playing around with it or using it…there needed to be a bit more top-down push to get people just to even experiment with it.”

Here are a few of the obstacles Kinloch said she identified.

Learning AI is a second job. Finance people, as you may have noticed, are busy. And according to Kinloch, her team was very focused on day-to-day work. Asking them to spend time learning a new tool was a challenge in that environment, and made adoption sluggish.

“It’s really hard to pick up a new tool and learn a different way of working without a top-down mandate and/or time to set aside to come up the learning curve,” she said.

When Richard Ranieri, an AI business consultant, works with a company, he combats this issue by performing an audit. He finds the biggest workflow bottlenecks and demonstrates how to solve them with AI.

“You find out the company’s biggest pain points,” he told CFO Brew. “Where they can use AI to get the best results, and how you can kind of work in tandem with people and AI so that it’s still a human-first solution, just using AI to help accelerate their situation.”

Failure is demonized. Kinloch encouraged her employees to use AI in one-on-one and team meetings. But encouragement can only go so far if people are afraid of failure.

Kinloch heard about an incident in which the finance team of another company dropped a board deck into ChatGPT and the commentary that came out was unintelligible to them.

“And the response was, don’t ever use ChatGPT again,” she said. “That’s actually part of the learning curve. You’re going to see failures first, then you can figure out how to tweak your usage.”

Tracking the wrong things. A widely publicized MIT study from earlier this year stated that most businesses implementing AI failed to see any revenue gains. But Ranieri believes tracking revenue to understand AI’s success or failure is a mistake most companies are making.

“Although their revenue might not have changed instantly, they were able to save 10 hours a week on repetitive tasks that were originally being done by a person,” he told us in an email. “Now that person has more time to start doing other tasks that are more beneficial to the company.”

Kinloch agreed that trying to isolate revenue as a KPI would be difficult, and probably wouldn’t capture the increase in incremental revenue.

Solutions for AI hesitancy. So how did Kinloch face these issues at her company? First, she tried to gamify the use of AI. She had employees demo their different use cases at a monthly meeting, and awarded giftcards for first, second, and third place. A little incentivization can go a long way.

Ranieri agreed with this method.

“Make it fun,” he said. “By keeping it engaging and bouncing ideas back and forth, having a dialogue, that definitely helps.”

Kinloch also saw that her junior employees—ones less steeped in the current processes—used AI better and adopted it more quickly.

But the big thing for both Kinloch and Ranieri is training. According to Ranieri, a lot of companies rush in and add tools without training the people or training the data. In fact, only 38% of companies that have implemented AI tools have offered employee training, according to the OwlLabs and Pulse survey. Kinloch also stressed the need to “set aside time,” because she believes “it’s unrealistic to expect people to spend their nights and weekends” learning AI.

“Humans are just used to what you’re used to. And so picking up new tools and ways of working is pretty difficult. Companies that have been most successful in deploying AI at large scale have really put a lot of time and effort into the change management process.”