Most companies that piloted AI over the past two years now face a harder question: why the tools aren’t delivering at scale. According to BambooHR’s Redesigning Work report, 63% of organizations have increased their AI tool budgets in response to employee usage. Yet workers spend more of their AI time troubleshooting and rewriting prompts than moving their work forward. The shortfall often traces back to how the tools were introduced. Many were layered onto existing processes without anyone asking whether those processes still made sense. For growing companies, getting that question right before rollout matters more than rolling out fast.
Allison Heintz is a veteran people executive and strategic advisor with more than 15 years of experience leading HR through growth, transformation, and disruption. She’s held senior people leadership roles at companies including Magic Leap and Cognizant, and has also advised CEOs, boards, and executive teams on operating models, organizational design, workforce strategy, and M&A integration. Her approach to AI follows the pattern she applies to any transformation: start with the business outcome, then diagnose what’s getting in the way.
“The way I look at it, AI is not a technology strategy. It’s a workforce strategy. It’s really about how we reimagine what work we’re doing,” she says. That view changes who needs to be at the table and which questions come first. Rather than asking where a tool fits, she starts by asking what the work is meant to accomplish.
Redesign before you deploy
In her conversations with companies, Heintz sees teams skipping the hardest step. “We’re trying to layer AI onto the pieces of the process that already exist, as opposed to thinking through the lens of the business strategy we’re really trying to accomplish and the capabilities that are needed,” she says. “Does that work even need to exist anymore? Should that work evolve into something completely different?”
The Redesigning Work report shows how far roles have already drifted from what’s on paper. Three in four workers say AI has meaningfully changed how they do their jobs, yet HR pros say they’ve updated fewer than half of their workforce’s job descriptions to reflect it. Heintz acknowledges that redesign is slower than adding a tool, and that the pull toward a quick answer is only human. “You can’t get away from the fact that it takes a bit of a slowdown to go fast,” she notes.
Give agents a job description
Measurement is where many rollouts stall. Heintz traces disappointing returns to a rush to put AI everywhere without checking whether it performs. Many companies let employees build their own agents, and those agents now run across the business with little oversight. “Agents, to me, need an ID. We need to be able to track them. Their job needs to be defined. They need a performance review,” she asserts.
The gap between adoption and oversight shows up in the data. Ninety-two percent of HR pros are piloting or planning AI agents across the HR lifecycle. Meanwhile, 54% of organizations lack a clear, documented AI usage policy that’s consistently communicated. Heintz sees the early enthusiasm as a normal first stage and a signal to get more deliberate. “We kind of threw it out there like spaghetti, and we’re seeing what sticks,” she says. “How about we now figure out what the recipe is? What ingredients do we really need? What are we even trying to make here?” She expects companies that can answer those questions, and tie the answers to business results, to pull ahead.
Tie ownership to outcomes
AI has also set off turf battles. “It’s a little bit like the cool new toy that everybody wants to say is theirs,” Heintz says. “I’m definitely seeing that friction point, with everybody wanting to own the strategy. But the reality is that it’s a workforce strategy, and it’s not necessarily owned by one person.” Executive alignment comes first, so the organization hears one consistent message about how the business plans to use AI. From there, she looks for the function best positioned to serve the company’s business strategy. That won’t always be technology, operations, or HR.
Her test for that function is how it defines success. “If there’s a team that measures outcomes by ‘check the box, we did it,’ that’s probably not the right team to be running your AI strategy,” Heintz explains. The right owner works backward from the result the company needs and builds an execution roadmap measured against it. “If it’s a trend out there that I can’t connect to business performance, I’m not really interested,” she adds.
Fix the system, not the people
As rollouts pile up, fatigue follows. Heintz expects some of it, since the change is fast and ongoing, but she looks beyond individual employees for the cause. “A lot of people are talking about fatigue through the perspective of change resistance, and they’re making it a people problem. I think that’s too convenient,” she says. “People resist when the case for change isn’t clear, when the new behavior is harder than the old behavior, or when the systems around them reward the old way of working.”
Her remedy starts with communication that goes beyond announcements. “Double down on your communications, and not just sending out emails, but really think about how you connect people to the work and the outcomes you’re trying to achieve,” Heintz notes. It extends to the environment people work in, because AI changes the whole system rather than one piece of it. “If you want a different outcome and a different behavior, you really need to redesign the environment.”
Know when to pump the brakes
For small HR teams, every rollout competes for the same limited time and budget. Heintz’s warning sign is simple. “If you can’t say what you’re trying to achieve and how that outcome helps business performance, pump the brakes right there,” she advises. Productivity gains alone don’t clear her bar. The goal is a defined business result, whether that’s performance, customer experience, or whatever carries the most value for that organization.
“If you can’t connect it to that, then there’s a question you want to probe a little deeper into before taking the time, and sometimes the dollars, to invest in this technology,” Heintz concludes.
