AI Rollouts Live or Die on Whether People Actually Change How They Work

Companies keep treating AI adoption as a technology rollout, budgeting for tools and licenses and pilots, then acting surprised when usage stalls. The systems install fine. What doesn't install is the new behavior the tools depend on. The organizations getting real value out of AI are treating it as a change management problem first, making the case for why the work is changing, watching whether people actually work differently, and holding managers accountable for reinforcing it.

Ann Marvin is president and CEO of Peacock Hill Consulting, powered by IMA Worldwide, where she helps organizations move technology initiatives past installation and into sustained use. A technologist by background, she's spent years watching the same pattern separate the initiatives that stick from the ones that stall out, and her read on AI is that the technology has outrun most organizations' ability to absorb it. The tools are ready. The people are moving through the change at very different speeds. "You're set up for the situation you already have, because that's what your reinforcements are backing," says Marvin.

Two paths through the same change

Marvin sees every AI rollout running along two connected paths at once. One is personal, where an individual employee lands somewhere on an adoption curve that can shift by the day. "You've got that personal path where the employee is out there loving it, running it, using it," she says. "And you've got the other path where the person's out there hating it, afraid to use it, doesn't want to touch it."

The other path is organizational, where sponsors and leaders set out what they're trying to accomplish. The two rarely move in sync. The organizational message usually arrives as some version of cost savings and automation, and it collides with the personal question every employee is already asking about what the change means for them. "Even if the organization is on board, they're heading back to that personal path, and they're struggling with the employees both ways," she says.

Nobody moves without knowing why

The failure Marvin sees most often has nothing to do with the tooling. It's that people were never told why the change is happening. "The whole thing fails when people don't understand why they're doing this," she says. "They don't understand what it is. And then you get into, what's in it for me? What does it mean for me?" Her firm opens engagements with an implementation history assessment, built on the premise that an organization will keep doing what it has always done unless it deliberately decides otherwise, and that culture beats intention nearly every time.

AI raises the cost of skipping that step, because it moves faster than the rollouts that came before it and exposes the gap sooner. "AI speeds up the process, so you run into that brick wall quicker than we have in the past," she says, which turns a soft foundation into a fast failure the moment work goes out without the groundwork.

Behavior first, results later

When Marvin's team defines success for a change, the metric isn't only the business outcome. It's the specific behavior they expect to see people adopt, because that shift is visible long before the financials move. "The business value is a lagging indicator, but you can see someone changing their behavior up front," she says. The lever that moves behavior is reinforcement, and she puts it above communication and training.

"People react based on what leaders reinforce," she says. "It's got about three times the impact of saying it or doing it." When the behavior a change requires runs against how a company actually operates, she frames the choice starkly. "You have to change the change or change the culture, and changing the culture is huge." Most of the durable advantage, in her account, sits on the human side of the work rather than the platform.

The middle layer or the black hole

Reinforcement only works if it cascades, and the place it breaks is the middle. Senior leaders often route around their managers, going straight to frontline staff or never setting up the reinforcement with their own direct reports, and the layer in between goes dark. "If you skip the middle layer, they're not part of those conversations," she says. "If their manager doesn't have this conversation to reinforce them, it breaks, and that's where you get that disengagement we call a black hole."

Her fix is to stop treating middle managers only as enforcers and start treating them as people moving through the same change, with the same questions about whether they're willing, informed, able, and in control of what's being asked.

Tools versus what you're trying to accomplish

Asked what will separate genuinely enabled organizations from the ones stacking up licenses over the next few years, Marvin points back to the conversation most companies skip. "We almost throw tools at everything instead of thinking about what we're really trying to accomplish," she says. The organizations that pull ahead involve the people affected early enough that they bring back the solution themselves. "Are they having these conversations? Are they involving people? Are they listening to what they're saying back?"

The alternative is a company stalled early with people pushing in different directions. Her firm packages the alternative as a repeatable change-capability journey, and her framing of the moment is that the speed of the technology raises the bar rather than lowering it. "Agentic AI may accelerate the work," she says, "but leaders must still accelerate implementation."