What she is finding there is that the companies pulling ahead on AI are not distinguished mainly by their technology. At Fortune‘s AIQ Summit on Friday, David, who also carries the title of ServiceNow’s director of futures, said the widest gaps between the leaders and everyone else are about people.

“Talent investment. That was the biggest gap,” David told moderator Jeremy Kahn, Fortune‘s AI editor. In ServiceNow’s Enterprise AI Maturity Index, 57% of “pacesetters” invest in ongoing AI upskilling, versus 4% of other organizations. By the report’s figures, the widest single gap is in attracting, hiring and retaining AI talent: 68% vs. 10%.

“It’s really operational discipline, right?” David said. “It’s not just technology, operational discipline, and transformation.”

Pacesetters are the roughly 21% of surveyed organizations that scored above 60 on the 100-point index, averaging 74 against 45 for the rest. Their edge extends beyond talent. By David’s account, 64% are on a path to unified data, versus 14%. And around 57% have “a clear, strong AI vision,” versus 21%.

Then there are the autonomous workflows. About 59% of companies surveyed are past the pilot phase, David said. “Only 9% are doing agentic, autonomous, multi-step workflows that are not checked by human beings at every step.” Among pacesetters the share is 36%, against 2% of others. These workflows are “not just one little point of sale,” she said.

That distinction drives what David called her “pet peeve”: “AI adoption for everyone.”

“I get it, because it’s a great CEO story,” she said. “95% of everybody has X, Y, Z. But it is not translating into productivity, and it’s often not with training, and that doesn’t equal maturity.”

Kahn had set up the problem. Many companies gave employees AI tools and “let a thousand flowers bloom,” he said. Then they decided it was too expensive, with no productivity gain. Now they are narrowing to specific use cases. Kahn also pointed to companies that dropped agents into existing workflows, or swapped an employee for one.

David said the answer starts higher up. “I think it starts with leadership,” she said, “growth mindset, an idea, and a vision for change of what the company can become, followed with specific training around AI and specific functions.”

She urged executives to think “almost in a day zero mentality” about what process would be designed from scratch. The alternative, she said, is to “sprinkle AI on something probably a very old or broken process to make it go even faster. That’s just not solving the problem faster.”

Kahn pressed on a worry he said executives voice: the technology moves so fast that a redesigned process may be obsolete next year. David said she hears that too. Her answer was to have a vision, align the data, choose workflows, assign ownership and “iterate.” “I don’t think that any company is going to succeed by doing some kind of top-down plan where one person or one strategy team is redesigning everything immediately,” she said.

She pointed to ServiceNow’s AI Control Tower as one way to see across the organization. “You have to have transparency,” she said. “Having the visibility from across the organization and a very strong foundation allows you to be flexible.” ServiceNow produces the index, and David works for the company.

Most companies are behind on the human side. According to the report, 59% of organizations lack long-term HR plans for AI, and 42% of employees say they aren’t getting enough AI training. The report puts pacesetters’ average ROI at 160%.

The leaders are not confined to tech. David said technology companies took about one in five spots on Fortune’s AIQ 75, followed by financial services. Construction and engineering took six, and food and healthcare five each.

“It is everything, everywhere, all at once,” David said.

The index itself has been volatile. “In 2024, we started this, and we thought it’s just going to go up and up and up,” David said. Then the average score fell from 44 to 35. Kahn noted the drop came “several years into the generative AI boom.” David said the cause was that “the noise was out there, AI was everywhere, but it hadn’t really filtered in to company value.” Scores rebounded to 51 this year.

“I can’t promise you that there won’t be another dip next year,” she said.

Next year, she said, ServiceNow plans to add agentic AI to the AIQ, and the measure will turn on “what are we deploying and where,” and whether companies are “realizing actual commercial value.” That, she said, “is what I think true maturity for business is going to mean versus everybody has, you know, AI in their pocket.”

For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.