AI Strategy Has Become Corporate Theater. Here’s How CEOs Are Turning It Into Business Results.

AI is now the reason U.S. companies give most often for cutting jobs. Through August, employers attributed 116,175 announced layoffs to AI, according to outplacement firm Challenger, Gray & Christmas.

Inside companies, the story and the work don't always match. A June survey from BCG found that of nearly 12,000 workers, only 28% of frontline employees saw a strong connection between what leaders say and what the organization actually does in this era of digital transformation.

That gap shapes how companies announce layoffs, how investors react, and whether employees actually implement the tools in their workflows.

What the layoff numbers show

AI led every month from March through July, peaking in May at 38,579 cuts, or 40% of the total. In August, it fell to 3,462 cuts, about 7%, behind restructuring, market conditions and closings, as IBTimes UK reported. The drop could mean fewer AI-driven cuts, or companies describing their cuts differently.

Some executives now say outright that their cuts are not about AI. Microsoft Chief People Officer Amy Coleman wrote in July that about 4,800 eliminated roles "are not being replaced by AI," and Etsy CEO Kruti Patel Goyal told staff in August that about 220 job cuts "weren't driven by AI."

Others still lead with the technology. On Sept. 10, Booksy CEO Stefan Batory told LinkedIn followers that the beauty-booking platform would cut about 20% of its staff to pursue an "AI-first vision," without saying what the technology had changed. "The message doesn't state why these cutbacks were necessary," one commenter wrote.

"Naming AI in a layoff announcement can win over investors while pushing current and prospective employees away," said Andy Challenger, the outplacement firm's chief revenue officer.

Block's shares rose nearly 18% the day after the payments company said in February it would cut nearly half its staff, citing "intelligence tools." But a CNBC review of 23 S&P 500 companies that tied layoffs to AI found 13 trading below their announcement-day prices as of May 15, even as the broader market hit new highs.

In a Gartner survey of 350 executives at companies using AI agents and similar tools, about 80% reported workforce reductions, at nearly equal rates whether returns were higher, modest, or negative. "Workforce reductions may create budget room, but they do not create return," said Gartner analyst Helen Poitevin.

Measuring work, not logins

"Look at what the company measures," said Johnny Chang, CEO and co-founder of Modo, which tracks whether AI tools change how work gets done. "If the numbers are seats, logins, tokens, and training completed, none of that measures impact or outcomes. Those count tools, not work."

Chang told International Business Times about a sales operations manager customer who spent 6.5 hours every Monday merging four lead lists by hand. Once the task surfaced, "a short script did the merge in seconds, and the sales floor went from 200 new leads a week to 200 a day," Chang said. "The fix was easy. The hard part was seeing the work as it was actually done."

Before announcing AI gains, a company should know "which specific tasks changed, by how much, and which business number moved as a result," Chang said. Fewer seats or fewer people is "a forecast dressed up as a result."

Some companies selling AI tools now make their case in those terms. Upfirst AI, which builds AI phone receptionists for small businesses, frames the math in missed calls: according to the company, the value at risk for a single unanswered call sits at around $1,900 for a roofing company and around $620 for a legal practice. Altorney, a legaltech company designed to empower law firms, presents the results of its MARC solution as a benchmark against a full manual review, measured in cost and cycle time. And Sonata Software's Rajsekhar Datta Roy, the company's CEO, told CXO Media in January that each of its AI engagements is "anchored to specific business outcomes."

"If what gets reported is 'we cut headcount,' that doesn't measure what the AI did; it measures a business decision the company made afterward," said Hernan Fino, Head of Innovation at Making Sense, which runs AI adoption programs for mid-market and private equity-backed companies. Honest measures, Fino said, sit in the process: cycle time on a specific task, or automated decisions later reversed. That is "less impressive on a slide, and much harder to poke holes in during due diligence."

Finance has its own version of that test. "Headcount alone is not a measure of operational efficiency," said Shagun Malhotra, CEO and Founder of SkyStem, which makes month-end close and reconciliation software. "CFOs should look at indicators such as close time, reconciliation exceptions, manual intervention, and the consistency of financial reporting. These metrics show whether the underlying process is actually becoming more reliable and scalable."

The trust gap

Kirsten Moorefield, Head of Research at Cloverleaf Labs, the research arm of team development platform Cloverleaf, said the top reason people don't use AI "is because of a lack of assurance." When a company cuts jobs and points to AI, "this destroys trust."

The tools can make it worse, Moorefield said. When employees brought a real workplace conflict to an AI model, Cloverleaf Labs found, "less than 1% of the advice it gave was about repairing the relationship."

"Leaders must get in front of that by investing in a culture of genuine relationships," Moorefield said. "This means equipping managers day to day, as well as leadership doing what might be the scariest thing of all in market upheaval: being authentic."

A 90-day test

Fino's alternative to another strategy deck is to "start with a single workflow, not a vision."

The place to begin, he said, is a decision that keeps getting stuck: "the point where someone has to stop, pull information from three different systems, and decide with incomplete information."

AI pulls that information together, and the person still makes the call. "Then you measure: how long did that decision take before, how long it takes now, and how many exceptions got resolved without escalating," he added.

Chang's advice fits the same exercise: start from "how work happened before AI" and tie results to something "the team already reports, like leads per week or tickets resolved."

Malhotra would start by "examining where the finance team is spending the most time on repetitive manual work or resolving recurring exceptions."

She said that "before automating, you need to understand the process itself... Otherwise, you risk making an inefficient workflow faster without addressing the underlying problem."

All three depend on finding the friction, which Fino said is "visible if you ask the people who do the work." Managers must close that trust gap, Moorefield said. If not, "people will share their ideas less, which is the death of all agility and innovation."

"You don't transform a company in 90 days," Fino added. But a CEO can show "one real process, made faster, with clear before-and-after data."

"Only then does it make sense to ask where else it could apply."