Harvard Business School and Microsoft recently published a serious diagnosis of why AI transformation stalls. Drawing on a closed-door summit with a dozen global organizations, they name seven specific frictions: productivity gains that don't appear on the balance sheet, process debt that AI surfaces faster than organizations can resolve, and an identity challenge that goes deeper than reskilling. Their prescription: redesign processes from scratch, make role changes explicit. It's rigorous work, and the diagnosis is largely right.
But the pipes aren't the problem. The problem is the pulse.
The people inside the org chart
Most transformation programs think in processes and org charts. But the same shift is happening inside every individual in those charts simultaneously.
A senior specialist whose job used to require years of accumulated knowledge is watching a tool do a version of it in thirty seconds. That's not a reskilling challenge. As Johannes Kleske puts it: when expertise becomes cheap, and AI is making it cheap very fast, value moves to judgment and accountability. Their expertise didn't disappear. The scarcity that gave it meaning did.
They're not looking for better governance frameworks. They're looking for others asking the same questions.
If the transformation doesn't reach them, not as a rollout but as a genuine reckoning, you haven't transformed anything. You've just changed the tools.
The question underneath the urgency
Most executives I speak with aren't primarily worried. There's a question they're circling: What could this actually make possible?
When the average is automated, we are forced to find what is truly essential about our work. For the first time in decades, we have real conditions to build what management theory has promised for fifty years: teams that are genuinely self-directed, purpose-aligned, able to act without constant coordination overhead. AI absorbs the operational layer that used to keep people busy executing rather than thinking. That frees something up that process redesign alone never could.
That's not a footnote to the efficiency story. That's a different story entirely.
The thing I keep coming back to
A contract gets drafted by an AI agent in seconds. Then it sits in a legal review queue for two weeks.
That gets framed as a governance bottleneck. Add an oversight layer, redesign the review process, move on.
But that's not what's actually happening in teams.
Different parts of the organization are now operating at fundamentally different speeds. That doesn't just create bottlenecks. It creates friction in how people work together, how decisions get made, how trust forms, how meaning gets distributed.
The technology moves fast. Organizational culture moves slow. Individual learning moves at whatever pace each person can manage. The informal norms that actually govern behavior move slowest of all.
Teams are navigating between all of these rhythms simultaneously. Every day.
We keep calling it an adoption problem. A training problem. A communication problem.
I think it's a rhythm problem.
And rhythm problems don't get solved with oversight layers. They get solved by redesigning how people move together.
What changes if you look at it that way
If the real challenge is rhythm, the intervention looks different. You're not trying to get people to use the tools. You're trying to rebuild the rhythms that let a team function when the tools have already changed everything around them.
That's a harder problem. It's also a more interesting one.
The goal isn't to keep up with the machine. It's to remain the author of the story the tool is helping you write.