By Mark Maurer, Kristin Broughton and Jennifer Williams

Public companies frequently mention generative AI on earnings calls, citing its positive effect on the bottom line or promising results in tests. But there's one application they leave unspoken: the technology's role in those very calls.

Investor relations departments at companies such as shoe brand Skechers USA and networking-systems and software provider Ciena have begun using generative artificial-intelligence to help prepare their earnings commentary.

Some have used generative AI to predict the questions analysts might ask, for example, and to ready the best answers.

Finance teams in general have been slower to apply AI than other departments, like marketing and customer support, in part because they were still weighing the risks , costs and benefits. But they have begun to make an exception for investor relations.

Every quarter, U.S. public companies report their earnings with a rollout that involves a detailed earnings release and presentation, followed by an earnings call on which executives deliver prepared remarks and run a gantlet of analyst questions.

"Due to the language and communications involved in investor relations, generative AI has clear application for these processes," said Shannon Cole, vice president in the finance practice at research firm Gartner.

Indeed, 44% of investor-relations professionals have already folded AI into their companies' IR programs, primarily to create written content, according to a March survey from the National Investor Relations Institute, a professional group.

Some companies rely on AI as a way to gut-check their earnings approach. In recent quarters, Skechers has used an AI tool to see whether recent earnings commentary aligns with the brand's intended messaging, Chief Financial Officer John Vandemore said.

The move is aimed at gauging whether investors' AI tools -- which they may use to digest earnings reports instead of actually listening to calls or reading company materials -- would spit back the same key messages Skechers expects, he said.

"The reality is a lot of investors are going to gravitate to using tools like that," Vandemore said. "So we want to make sure as we prepare and deliver messages that if they go through that analysis ... what comes back is consistent with the objectives of the messaging we're attempting to deliver."

Ciena is using AI as "very strong assistants" that can help draft the script of an earnings call and anticipate questions, said Gregg Lampf, vice president of investor relations.

AI can identify word choices or phrases that listeners might not fully understand. "We can run ambiguity checks, for lack of a better way of putting it, based on certain personas to help ensure that we are, in fact, articulating something in a way that will make most sense to most people," Lampf said, referring to the different types of people who listen to earnings calls.

Executives at many companies are using AI to refine word choice in their prepared remarks, for instance, in deciding whether to say the quarter was "strong" or "solid," said Dan Sandberg, head of quantitative research and solutions at S&P Global Market Intelligence. The firm's tool recently preferred "strong," based on the earnings metrics of other companies that used the word on their earnings calls, he said.

Generative AI can also read the harmonics in executives' prepared statements on earnings, assessing them as upbeat, gloomy or something more measured, said Steve Soter, vice president at business-reporting software provider Workiva.

"Is the CFO worried or optimistic, and could that give some indication as to how future performance is going to look for the company?" Soter said, describing the sort of possibly unintended information that AI can catch before it goes out.

Increased reliance on AI in drafting news releases or conference-call scripts poses new risks for companies, advisers say, among them ethical issues, like transparency and accuracy. But the biggest may be keeping information private. Businesses that have adopted or invested in private large language models have better security for their not-yet-public details than those that turn to public tools, they said. Uploading private, material information to a public domain opens the possibility of unintended disclosure.

"It becomes very seductive to turn a little bit away from the risk factors, and just kind of lean in wholeheartedly because you just feel like, well, it's going to make your life a lot easier," said Evan Pondel, chief executive of investor-relations and strategic communications firm Triunfo Partners.

For CFOs, an AI assist with earnings-day prep could be a gateway to getting comfortable with the technology as they look for ways to clip costs, become more efficient and increase revenue, said May Habib, chief executive of enterprise AI startup Writer.

"IR is just such a great way for the CFO to hands-on understand what AI can do so that it doesn't feel like magic," Habib said.

Write to Mark Maurer at mark.maurer@wsj.com, Kristin Broughton at Kristin.Broughton@wsj.com and Jennifer Williams at jennifer.williams@wsj.com

(END) Dow Jones Newswires

November 19, 2024 06:00 ET (11:00 GMT)

Copyright (c) 2024 Dow Jones & Company, Inc.