As we close out the week, I find myself still thinking a lot about the public plenary sessions from our WSJ CFO Council Summit—our first on the West Coast—which attracted finance chiefs and tech experts from Palo Alto and beyond. These leaders talked about the promise of AI investments and some of the perils, including massive job cuts.

What many participants found invaluable, though, were insights gleaned from private conversations and our peer-to-peer Exchange sessions. Those are closed discussions, but some leaders were chosen to share top takeaways from their dialogues, where members talked about their experiences.

I noted that I started my career years ago as a bond reporter, covering credit markets and all the major rating agencies, and later structured finance and securitization (that pre-2008 was then seen as the latest financial accelerant and great innovation, which also was fraught with great promise and risk).

So I was pleased to connect with my co-host Noémie Heuland, CFO at Moody’s, for our exchange session on “The Economics of AI: Managing Cost and Capturing Value.”

Here are the three key takeaways that Noémie shared:

✈️ How to Move From Pilot Phase to Scale

Start with a clear “North Star” use case. Some examples of where to set AI goals could include areas such as analyst productivity or customer self-service.

Establish a gated delivery path: Move from sandbox → controlled pilot with human-in-the-loop → limited production → scaled rollout with monitoring (quality, drift, cost, risk).

One example from Moody’s that Noémie noted was how the rating company used a small central leadership team to roll out Gen AI access to all employees (which they called “14,000 innovators”) while a lean Generative Intelligence Group (GiG) vetted tools, enabled reuse, and protected security, trust and accuracy.

⚖️ How to Balance Managing Costs With Investing in Long-Term Planning and Innovation

View Gen AI as a “risk of inaction” problem.

• Key quote: “Standing still is viewed as riskier than moving fast, so we discussed the need to prioritize real business impact with strong governance over cautious experimentation,” she said.

Set baselines and counterfactuals. Measure the current process (cycle time, error rate, cost) and compare against AI-assisted workflows in controlled rollouts. If AI drafts a first research report and the expert edits, the value is analyst-hours saved and fewer rework cycles.

Track the “shape” of cost over time. Training/fine-tuning and integration are often front-loaded; inference costs scale with usage, she said.

👩🏽 v. 🤖 Finally, What Role Do Humans Still Play? (In other words, how do you manage and marry AI with Human Intelligence?)

Judgment on high-stakes decisions. The goal is not to replace human expertise but to amplify it. (Moody’s ratings committee, for example, is led by humans with deep institutional knowledge, history and 120-plus years of understanding the business through dialogue with issuers, she noted.)

Context, nuance and "ground truth." The qualitative judgments, forward-looking opinions and analytical reports produced by human experts serve as a critical labeling, validation and enrichment layer for AI models, she noted. The machine learns patterns, but humans provide meaning.

Gen AI is not replacing human expertise but augmenting it. AI allows teams to focus on high-value decision-making while AI handles the data-heavy lifting.

Editor's Note: I’m curious if readers agree with these points, or have other reactions. If so, please reach out to me directly by clicking Reply to this newsletter.

And read on below for more highlights, and takeaways from our CPO Council Summit, also our first for the Chief People Officer community.

—Walden Siew
Bureau Chief, CFO Journal

📆 Earnings (Short and sweet today)

• Carnival

The list of layoffs continues to mount, this time at Sheryl Sandberg’s Lean In nonprofit…

The organization has shed about a quarter of its staff as the founder focuses the feminist organization on pushing back against the “tradwife” and manosphere movements that prioritize women’s roles as housewives, Alexander Saeedy and Emily Glazer report.

The numbers, and a new CEO. More than a dozen employees have left the Sandberg Goldberg Bernthal Family Foundation, which includes the LeanIn.Org initiative, over the past year, either through layoffs or by choice, people familiar with the changes said.

On Tuesday, Sandberg announced that Bridget Griswold, 25, was taking over as chief executive of Lean In, pointing to her background in artificial intelligence. Griswold, a 2023 Brown University graduate, spent two years as a Meta product manager before her elevation. Her rapid ascent—and lack of nonprofit experience—have led some staff to leave, the people said.

What does Aaron Levie, Box co-founder and CEO, think about the future of work?

Here’s Levie talking about how the software company and their leaders think about the use of AI agents to boost efficiencies.

“So we were able to automate 5% of a company's operations. We can automate the ERP operation. We can automate managing workflows around our CRM, or IT workflows, but actually most of work is this sort of messy, unstructured human collaborative—lots of different kind of data types coming in—and we've never been able to automate any of that,” he said, speaking at our inaugural CPO Council Summit in Palo Alto, Calif. 

“And so if you just did a heat map of all the work that happens in the world, the vast, vast, vast majority, probably like 90, 95% of the work that we do, is inherently unstructured, and for the first time ever AI allows us to bring automation to that part of the work.”

Is it bigger than previous waves of automation?

“Easily by orders of magnitude, and it sort of certainly deserves its own way of thinking about what that future of the information age is going to look like. It's actually interesting we probably called the information age too early.”

On ways to use AI:

“The way that we think about AI is, essentially 90% of the way that we think about it, is we want to use it to do more things.”

“We want to use it to ship more code faster, better marketing campaigns, serve our customers better even in areas where we are going to save money,” Levie said. “So we look at every opportunity where we have automation, and we say can we automate something to free up dollars to go and actually do the thing that we couldn't have done before.”

📰 Other headlines

• Jeep Cherokee Production Halted Over Supplier Dispute
• Court Dismisses X Lawsuit Alleging Brands Illegally Boycotted the Platform
• FCC Advances Effort to Bring Telecom Call Centers Back to the U.S.
• Exclusive: The SpaceX IPO Will Be Just as Unconventional as Musk Himself
• Anthropic Wins Injunction in Court Battle With Trump Administration
• The Private-Credit Industry’s Trouble: Surging Redemptions, Slower Fundraising

📈 Earnings wrapup

• Arms Maker CSG Eyes Deals in Push to Boost Production, CFO Says
• Designer Brands Comparable Sales Fall
• Polish Fashion Retailer LPP Shares Jump to Record on Higher Sales
• Meituan Posts Quarterly Loss as Price War Continues

For more earnings news, click here.

The Wall Street Journal's CFO Journal offers corporate leaders and professionals CFO analysis, advice and commentary to make informed decisions. We cover topics including corporate tax, accounting, regulation, capital markets, management and strategy.

Follow us on X @WSJCFO. The WSJ CFO Journal Team comprises reporters Kristin Broughton, Mark Maurer and Jennifer Williams, and Bureau Chief Walden Siew.

You can reach us by replying to any newsletter, or email Walden at walden.siew@wsj.com.