The Seams

You’ve been asked to use Claude. So has everyone else. That’s how it starts in most organizations right now. A tool arrives. A memo goes out. Training gets assigned. And within weeks, two things happen simultaneously: some people use it, some don’t. The ones who do get better. The ones who don’t get worried. And your organization has just sorted itself into two groups that BetterUp’s research calls Pilots and Passengers.^[1]

The problem is not the tool. The problem is that “we have a tool” is not a transformation narrative. It’s a deployment narrative. And deployment narratives fail at the seams—those invisible spaces between what the product promises, what organizational change requires, and what your people actually need to adapt.

The Belief That Shapes Everything

Here’s what research from BetterUp shows: a third of your employees already believe your real motivation in adopting AI is to replace them. Another 62 percent believe it’s to augment their skills. That gap matters, because your people will interpret every signal—every announcement, every workflow change, every decision about who gets trained and who doesn’t—as evidence for one of those two futures.^[1]

And the scary part: whichever belief takes hold calcifies. Organizations on the automation path see predictable drops in engagement, rising workload, and talent flight. Organizations on the augmentation path see the opposite: sustained engagement, capability development, and teams that actually build genuine skill with the tools.

The gap between those two paths is not created by the tool. It’s created by leadership signals. What managers say in 1-on-1s. Which workflows get redesigned and which get automated. Who gets a voice in how adoption happens. Those signals reach employees long before any formal communication plan does, and they set the terms for how people engage with the entire transformation.

If your people believe in your leadership’s intent to invest in them, they’re 46 percent more likely to believe you’re using AI to augment rather than automate. If they don’t believe that, no training program will shift it.^[1]

The Three Disciplines That Rarely Meet

This is where most transformations hit a wall. AI workforce transformation actually needs three distinct disciplines.

When those three disciplines fail to sit together, the seams widen. The result: AI gets dropped into 50-year-old workflows instead of being a catalyst to redesign them. Pilots emerge who figure out how to work differently. Passengers follow the old process, just slower. And your organization has created two different versions of work inside itself.

What Leaders at the Frontier Are Doing Differently

The organizations that are actually scaling AI—not just piloting it, but integrating it into how work happens—are doing something specific. They’re starting not with the tool, but with the work itself.

Katy George at Microsoft built a diagnostic that separates organizations that are moving forward from those stuck in pilots.^[2] It starts with a question most leadership teams cannot answer: Can we actually see the work AI is supposed to transform? If you cannot map your workflows task by task and role by role before you deploy anything, the AI will land on invisible work and produce invisible results. That visibility is the starting point. It sounds basic. It is not common. If you are not confident your organization has that picture, a new generation of providers is now using AI agents to close exactly that gap: running parallel interviews across entire teams to map actual workflows, tools, handoffs, and bottlenecks—capturing how work really happens rather than how leadership assumes it does.^[4]

From there, the frontier leaders ask the next structural question: Where can AI act, and where must a human decide? Not theoretically. Specifically. At the workflow level. If your organization is operating on assumption here—if there’s ambiguity about what decisions humans must hold—that ambiguity will kill your speed and calcify confusion about what the technology is meant to do.^[3]

The third move is this: go first. Publicly. Imperfectly. A leader at the frontier demonstrates their own AI use in front of their team. They share where it broke. They show real struggle, not mastered competence. That signal is worth more than a hundred emails about the strategic rationale. It says: we’re figuring this out together, not commanding you to follow.^[3]

The Middle Manager Problem Nobody Is Solving

Here is what the C-suite conversation misses almost entirely: middle managers are the critical layer, and they are being asked to carry the most weight with the least support.

BetterUp’s research identifies managers as the single biggest lever in AI adoption outcomes.^[1] Not tools. Not training budgets. Not executive mandates. Managers—because they set the team-level interpretation of every leadership signal. They decide, in practice, whether AI adoption is a threat or an opportunity. They shape whether their team members experiment or retreat.

But most AI transformation programs design for the extremes. They give executives the vision work and frontline employees the training. Managers get a deck to cascade and a mandate to enforce compliance. That is not a change management strategy. That is a signal that they don't realize what true change requires.

What the middle management layer actually needs is different: permission to learn out loud, a clear answer to “what does this mean for my team’s work,” and a protected space to redesign workflows at the unit level before those workflows get handed back up as precedents.

The harder point is this: being trained on a tool is not the same as having reorganized your work around it. A manager can complete every assigned module and still manage exactly as they did before. Real adoption happens when the manager starts asking different questions in team meetings—not “did you use the AI?” but “what did you do with what the AI produced?” That shift from usage to judgment is where transformation actually lands. It requires coaching, not compliance. And right now, most organizations are delivering compliance infrastructure and calling it a change program.^[1]

What to Stop Doing

Transformation demands subtraction as much as addition. A few things worth putting down.

Stop treating pilot utilization as a proxy for progress. A 70 percent usage rate in a contained pilot tells you the tool works in a controlled environment. It tells you nothing about whether work has changed. Measure workflow redesign, not login frequency.

Stop adding AI to existing processes without redesigning the process itself. AI laid on top of a broken workflow produces a faster broken workflow. The deployment is only as good as the redesign that precedes it.

Stop waiting for bottom-up adoption to carry the transformation. It won’t. BetterUp’s research is clear: top-down signal—leaders going first, visibly and imperfectly—is the condition that makes bottom-up adoption possible.^[1] The frontier firm research at Microsoft confirms this: the organizations moving fastest are the ones where senior leaders model the behavior, not mandate it.^[3]

Stop confusing access with capability. Giving everyone a license is a starting condition. It is not a strategy.

The hardest thing to stop is the narrative that this is going well because a pilot succeeded and no one complained. Silence is not adoption. It is waiting.

The Mindset Leaders Need Right Now

AI workforce transformation is not a technology problem. It is a sensing and responding problem.

The pace of change is fast enough that your six-month planning cycles will lag reality. Your organization needs to develop the muscle to detect weak signals—a task that quietly got automated, a team whose scope shifted, a hiring plan that slowed—and respond in weeks, not quarters. That requires two things: first, clarity about which signals matter for your business. Second, pre-agreed responses so that when the signal arrives, you do not spend three months in committee deciding what to do.

This is where the Methuselah Principle enters: durability comes not from specializing for the current moment, but from maintaining multiple capabilities in parallel across an uncertain future. The organization that optimizes entirely for AI-augmented work will be brittle when the technology shifts. The organization that keeps adaptability baked in—that runs production and experimentation simultaneously, that protects the humans who ask hard questions about AI alongside the humans who deploy it, that maintains optionality across its workforce—that organization survives.

And here’s the part that matters most: adaptability is a human capability, not a technological one. It lives in your leaders’ ability to hold two futures at once. It lives in your managers’ capacity to have the difficult conversation with a direct report about what AI means for their role. It lives in your organization’s willingness to let frontline people shape how adoption happens on their teams, not just comply with how it was designed.

Your Starting Point

The most honest diagnostic starts with a walk through the actual landscape. Not a workshop or a framework session, but a half day on real ground, with the people who know where your organization’s seams actually are.

The goal: to see clearly where your AI initiative meets the workforce reality it needs to work inside. To name the gap between what’s been announced and what people actually believe. To map which of your people are becoming Pilots, which are becoming Passengers, and what would actually move that ratio.

That reflection becomes your foundation for a 90-day roadmap that is not about deployment speed, but about structural change. It names where the seams are. It clarifies which human capabilities matter most in your specific context. It surfaces the leadership moves that create belief in augmentation over automation.

If this resonates with where you are, the fastest way forward is a direct conversation. Not a pitch deck, no generic playbook. An exchange about what you see happening, what you need to move forward, and whether the perspective here actually fits your reality.