Change management is no longer enough. AI can be a game changer, if we use it the right way.

The persistent illusion: change management is about designing a plan, then rolling it out 

In 2026, most transformations still fail, not because of poor strategy or inadequate tools, but because change management still rests on the illusion that you can define a trajectory and simply execute it. 

In a world of constant disruption (or at the very least one perceived that way), clinging to that illusion is no longer acceptable. In the age of AI, companies now have the means to continuously observe, adjust, and act. And yet, in most transformation programs, strategy is still designed upfront and then “deployed.” On paper, everything looks great: objectives are clear, plans are structured, milestones are defined. 

The problem is neither the strategy nor the execution. It is the disconnection between the two, and the absence of feedback loops. 

The reality looks very different. Take John, a key account sales executive who is expected to shift from a volume-driven approach to a margin-focused one, with tighter pipeline management and more structured client interactions. In theory, the goals are straightforward: improve conversion rates, shorten sales cycles, and prioritize high-potential opportunities. Training has been delivered, playbooks have been shared, and messages have been communicated. Yet in his day-to-day work, the change requires immediate effort. So he postpones, adapts, works around it, falls back into old habits, and gradually disengages. 

Sophie, who leads the change effort, either does not see it, or sees it too late. She keeps executing her plan, sending messages, and running rituals that grow increasingly disconnected from what is happening on the ground. The result is simple: the issue remains unresolved, and credibility starts to erode. 

This is how plans struggle to translate into reality. Even when the strategy is crystal clear, execution often depends on a few key individuals, the change ambassadors, sometimes poorly identified and always overstretched. Field signals don’t make it up the chain, or they do, too late, and adjustments happen slowly. The entire effort ends up relying on a handful of people instead of a robust operating model. Change becomes fragile, harder to predict, and execution turns fragmented, uncertain, and largely unmanaged. 

The classic mistake: using AI to accelerate what is already broken 

This is usually the moment when AI is expected to fix everything. And this is exactly where many organizations get it wrong. If Sophie uses AI to generate content, automate communications, or produce plans more quickly, she may solve nothing at all. Worse, she may amplify what is already flawed: generic messages get industrialized, poorly targeted actions scale faster, and approaches already disconnected from reality become even more so. In the end, Julien is overwhelmed with messages that do not reflect his day-to-day challenges, which is precisely the opposite of the intended outcome. 

Counterintuitive as it may seem, the real risk is not doing too little with AI. It is doing…too much, in the wrong direction. 

What AI finally makes possible: observe, understand, act 

The real opportunity lies elsewhere. It lies in AI’s ability to continuously reconnect strategy with operational reality. 

First, AI makes it possible to see what is actually happening. In Julien’s case, that means identifying that he still qualifies opportunities poorly, spends too little time understanding customer needs, and focuses on quick but low-margin deals. He may be applying some of the new standards, but not embracing the logic behind them. Capturing and interpreting these signals, something that used to happen too late or too broadly, becomes far more tractable with AI. 

Second, AI helps explain those behaviors. Julien is not resisting change. He simply does not see an immediate benefit. When the perceived effort outweighs the short-term gain, sticking to routine feels easier. That changes what the response should look like, Sophie no longer needs to send another generic message. She needs to trigger the right action at the right time. That could mean a recommendation embedded directly into the sales management tools when Julien qualifies an opportunity, or an automated prompt suggesting a key question to ask the client. In some cases, a simple comparison can instantly show him how deeper qualification could improve his chances of success. At the same time, his manager receives a focused, relevant, actionable signal. 

Used properly, AI enables organizations to leverage nudges and behavioral influence mechanisms far more effectively. The conversation shifts from broad change campaigns to fine-tuned adjustments, embedded in the reality of daily work. 

From plan-driven change to continuous steering 

This leads to the most structural shift of all: every action feeds a learning loop. If Julien changes his behavior, the approach is reinforced. If he does not, it is adjusted. Some levers are abandoned, others emerge, and priorities evolve. Change is no longer fixed: it adapts. That is the real turning point. 

The goal is no longer to define a plan and push it through. The goal is to set a direction, observe what is actually happening, and continuously adjust. In other words, to move from a design logic to an adaptation logic: less energy spent perfecting the plan upfront, and more energy invested in execution, execution that, in turn, constantly reshapes the plan through increasingly shorter feedback loops. 

Julien is no longer subjected to communications that feel disconnected from his concerns. Change becomes part of his daily workflow, appearing when the need arises, through levers that match his real usage patterns. It becomes less visible, but far more useful. 

Change leaders: stop rolling out plans. AI finally lets you adapt. 

In this model, Sophie does not disappear, but her role changes fundamentally. She is no longer there simply to produce plans or run training sessions. Her role is to provide clear direction, use automatically generated data and signals to understand what is really happening on the ground, and make rapid course corrections. She becomes less focused on content production and far more focused on understanding live dynamics in order to act quickly and effectively. Less planning. Less dependence on one or two key individuals. More decisions grounded in concrete signals. More real-time steering. 

The real risk is not that AI will replace the change manager. It is that AI will be used to industrialize outdated approaches. The challenge, then, is not to add AI to change management. It is to rethink change management for what it now needs to become: a continuous capacity to adapt, grounded in how people actually work.