Change Management
It’s Time to Move Beyond the Center of Excellence
Artificial intelligence is no longer a pilot program or a future-state aspiration. People aren’t waiting for a strategy, a governance plan, or an official green light. They’re already working with AI. Gallup found that 46% of employees were using AI at least a few times a year by the end of 2025, up from just 17% in mid-2023. People didn’t wait for permission. They moved. And now organizations are scrambling to catch up to something that already has momentum.
In the PATH framework — Present, Assess, Transition, Harness — this is Transition work. Transition is the space between what was and what will be. It’s where awareness catches up to reality. It’s where we stop observing change and start participating in it. Transition doesn’t ask whether we’re ready. It asks whether we’re willing to rethink, to participate, and to let go of frameworks that no longer match the world unfolding around us.
We are rocketing forward on technologies that few fully understand, trusting that someone must be paying attention. But in many cases, they’re not. Our systems of governance haven’t evolved to match the pace or scale of innovation. Part of why so many people feel like they’re riding blind right now isn’t a lack of intelligence or effort. It’s a mismatch between reality and mindset. The world has changed faster than our internal frameworks for making sense of it. And that gap, between where the technology is and where human understanding, governance, and culture are, is where anxiety takes root.
The Gap Is Not About Intelligence
Fear is a rational response to uncertainty, so the fear itself isn’t the problem. The danger is clinging to a mindset that no longer matches the reality unfolding around us. The questions this moment demands are not simply operational. They are ethical, civic, and human. What kind of future are we actually trying to build? What values should guide the tools we are creating? How do we make sure these systems serve humanity, not just efficiency, speed, or profit? These questions touch power, consent, responsibility, and dignity. And that means they don’t belong exclusively to engineers, executives, or policymakers. They belong to everyone who will live with the consequences.
History shows what happens when societies move fast without steering. The cost is always paid later, by people who were never invited into the conversation in the first place. We saw it in the consequences of the Industrial Revolution, in the fallout from the 2008 financial crash, and in the unchecked scaling of social media platforms that optimized for engagement before anyone fully understood the damage to trust, mental health, and public discourse. The lesson isn’t that innovation is dangerous. The lesson is that innovation without participation eventually creates distrust. AI adoption is following a similar trajectory inside many organizations right now.
The Hidden Risk Inside Centers of Excellence
Many organizations are still approaching AI adoption as if it were a traditional software rollout: build a strategy, select the tools, create governance, train the workforce, and measure adoption. The current trend is to formalize that instinct by building AI Centers of Excellence. The intention is understandable. Companies want governance, consistency, standards, and responsible implementation. But there’s a hidden risk in treating transformation as something managed primarily by a small group of experts, and that risk is the creation of the very disconnect organizations are trying to solve.
Microsoft’s own Azure Cloud Adoption Framework acknowledges that a centralized CoE approach works best as an early-stage scaffold. Over time, they say, organizations should transition to an advisory model and distribute AI expertise into product teams, platform teams, and the teams doing the actual work. Centralization that doesn’t evolve becomes a bottleneck, not a bridge.
When adoption is managed from a command center, a small number of people become the interpreters of the future while everyone else waits to be told what tools matter, what workflows will change, what policies apply, what skills they need, and what their role will become. That approach may create compliance, but compliance isn’t the same thing as trust.
This Is Not a Technology Problem
That distinction between compliance and trust points to something more fundamental. This isn’t simply a technology implementation problem. It’s a human transition problem.
Much of today’s resistance to AI comes from the same place as resistance movements throughout history. Not from ignorance, but from exhaustion with inevitability narratives that remove agency. It echoes something older and more familiar, something that looks like the Luddites of the 18th and 19th centuries rioting against machines that were dismantling their livelihoods, or agricultural workers in the Swing Riots attacking threshing machines not because they feared progress but because they were being told that massive changes to their contribution and dignity would happen to them, not with them. That same dynamic exists today. Many employees aren’t resisting AI because they hate innovation. They’re reacting to the feeling that decisions shaping their future are happening somewhere far away, made by people who won’t personally bear the consequences. And unintentionally, many organizations reinforce that feeling through the very way they’ve structured adoption.
Why Trust Is the Real Adoption Challenge
A February 2026 study out of a global consulting firm found that psychological safety is one of the strongest predictors of whether employees actually adopt AI tools. Not training, not access, not mandate. The determining factor is safety, meaning the feeling that it’s okay to try, ask questions, experiment, and learn without consequence. This finding held consistently across experience levels, role levels, and geographic regions. HRD Connect’s research on AI in HR for 2026 reinforces the same conclusion from a different angle: adoption depends on trust, clarity, psychological safety, and thoughtful change leadership, with governance and fairness shaping whether momentum builds or stalls. You can’t mandate your way to that.
What a Learning Ecosystem Actually Looks Like
What’s required is a more participatory model, not a command center but a learning ecosystem. The framework I developed here was inspired by a model I came across at one large enterprise that had used a similar structure for decades to onboard and nurture employees throughout their careers. The same logic applies to AI adoption, because learning how to work alongside AI is not a one-time training event. It’s an ongoing process that requires the same kind of sustained, human-centered support that organization figured out long ago. That model is built around three interconnected roles that together create the conditions for real adoption.
The first role is the buddy. Buddies are peers at the same stage of the learning curve, figuring it out together rather than waiting for someone with all the answers to show them the way. When someone navigates new territory alongside a colleague rather than alone, fear drops, imposter syndrome loses its grip, and agency returns. The learning becomes something that happens with another person rather than something being done to them. This is how people move from passive recipients of change to active participants in it.
The second role is the coach. Most people begin their relationship with AI the same way. They ask questions, copy prompts, generate some content, and walk away, concluding that AI is just average. That’s not a failure of intelligence. It’s a failure of context. The shift that changes everything isn’t finding better prompts. It’s when people stop asking AI for answers and start treating it as part of how work actually gets done, as a workflow layer, a cognitive support layer, and an active participant in the operating system of their day. That shift doesn’t happen through a mandate or a training module. It happens through experience, and that’s exactly what coaches create. They lead workshops and webinars, help teams redesign workflows around what AI can actually do, surface the friction points that don’t show up in executive dashboards, and connect day-to-day experimentation to outcomes the organization actually cares about. The question most people start with is “what’s the best prompt?” but the question that actually drives results is “how do I make AI part of how we work?” Moving teams from the first question to the second requires someone who understands both the technology and the human side of workflow redesign. Deloitte’s 2026 Global Human Capital Trends research supports exactly this kind of intervention: organizations that intentionally redesign roles, workflows, and decision-making to support human-AI collaboration are twice as likely to exceed their AI ROI expectations. Only 6% of leaders say they’re making that kind of progress. Coaches are the people who close that gap by making redesign practical rather than theoretical.
The third role is the mentor. Mentors operate at the level of governance and culture. They take what’s being learned across the organization and help translate it into norms, ethical guardrails, and policies that reflect the actual experience of the workforce, not just the priorities of the executive team. This matters because real governance doesn’t begin with policies written in isolation. It begins with shared understanding, and shared understanding only happens when people doing the work are invited into the conversation early enough to shape the outcome.
Together, these three roles form a system. Buddies surface what’s possible. Coaches build the capability to act on it. Mentors shape the norms that make it sustainable. No single role accomplishes what all three do in combination, and no centralized team of experts can replicate what happens when that structure is embedded throughout an organization.
The Future Belongs to Those Who Participate
People adapt faster when they feel they still possess agency within the process. They contribute more when their voices matter. They trust more when governance evolves alongside participation instead of being delivered from above as a finalized system.
Participation doesn’t require mastery of the math, the code, or the science. It requires curiosity, critical thinking, and the willingness to ask who this serves and who bears the risk. Real governance doesn’t begin with rules written in haste by people insulated from the consequences. It begins with shared understanding. And understanding begins when people are willing to sit with uncomfortable questions instead of outsourcing them to experts, lobbyists, or algorithms.
The organizations that get this right won’t necessarily be the ones with the most advanced tools or the most sophisticated Centers of Excellence. They’ll be the ones who understood, early enough to matter, that the future of AI at work is a human transition problem first. And they’ll be the ones who built structures to match.
