Nobody Owns the Decision

Here is a number that should stop a boardroom cold, and it is not the one with twelve zeros.

In Pearl Meyer’s Q2 2026 survey of 116 board members, chief executives, C-suite officers, and the senior managers beneath them, only about a third of the C-suite — 34% — said it is consistently clear which executive or team makes the calls on AI. Among board members the figure rose to 53%. Among the managers and professionals below the C-suite, it rose higher still, to 57%.1

Read that in the right direction, because the direction is the whole point. The people doing the actual work — the ones blocking and tackling on the ground — are the most likely to say someone clearly owns the AI decisions. The people at the top are the least likely. Ownership clarity falls as you climb. The closer you get to the seat that is supposed to hold the authority, the blurrier the authority becomes.

That is not a survey artifact. It is the seam, measured — the one this site has been describing since it launched, finally showing up in someone else’s data. The operator can see the seam because the operator lives against it. The executive cannot see it because the executive is standing on it.

The trillion-dollar symptom

The backdrop is what makes the number acute. At that altitude of spend, a chief executive knows the seat is at risk if the company falls behind its competitors or fails to deliver at home.

So the money is not the question. The money is committed. The question the survey exposes is quieter and more dangerous: all of that capital is flowing toward an outcome that no single, nameable person has been made to own. Everyone agrees the spend is real. A majority of the people at the top cannot agree on who answers for it.

Why the inversion is the tell

Pearl Meyer’s own reading, from principal Brad Jayne, is that AI did not create this. The ownership split is the symptom of a problem that has been sitting inside C-suites for years: leaders who each perform well alone and have never built the muscle of working as a team. The message from the top, he says, is often just start using it — and it skips the rest of the story, which is that leadership is not sure where to use it.3

A confused authority structure survived at human speed. AI collapses the hallway where it used to get sorted out.

That is exactly right, and it is worth being precise about why AI is the thing that finally forced it into the open. Every earlier wave of technology announced its own failure. A platform bought and never adopted sits there visibly idle; you can point at the unused license and the story tells itself. AI does not sit idle. It flows into whatever intention is already running. Erik Brynjolfsson calls this amplified intention,4 and readers here have met the idea before: if the operating intention of your organization is preserve what we have, AI amplifies preservation. Faster and more expensively than before, and without ever once looking unused.

The same property holds for ownership. A confused authority structure used to be survivable, because decisions moved slowly enough that the confusion got resolved in the hallway before anything shipped. Collapse the loop with AI and the hallway disappears. The loop moved: the decision now runs at machine speed, which means the authority for that decision has to exist before it runs, not get sorted out afterward. When it does not exist — when two-thirds of the C-suite cannot say who owns the call — the machine runs anyway, on whatever intention was already there.

That is the seam. Not a gap in the technology. A gap between what the organization permits and who is accountable for the permitting — the boundary between what an agent may propose and what the organization will allow, standing exposed and unowned.

You cannot bolt digital onto analog

This is the moment the survey stops being a headline and becomes a diagnosis. The reason “just start using it” produces a fractured, unowned result is not that the executives are careless. It is that they are treating an org-design problem as a technology purchase — bolting a digital capability onto an analog organization and expecting it to keep better time.

That is the argument The Back of the Watch makes at the level of the machine itself. You cannot turn an analog watch into a digital one by swapping parts; they are two different kinds of machine, and you do not retrofit from one to the other — you design the second and build it. An organization is the analog machine: authority routed through people, decisions resolved in hallways, a structure that assumes a human pace. AI is digital, and it runs at digital speed. Bolt the second onto the first and the mismatch does not average out. The analog structure stays exactly as analog as it was, now driving a capability it was never built to govern.

Automation does not fix a broken process. It accelerates one.

You cannot hide organizational failure at AI speed — when the constraint is where authority sits, speeding up everything around it just arrives faster at the decision no one owns. The Pearl Meyer number is that constraint made visible: not a slow step in the work, but an unowned one in the org. You can buy every model on the shelf and route it into every function, and if the authority for the decisions was never redesigned, all you have built is a faster path to the same unowned call.

Prior technology sat far enough from the authority structure that a broken one did not block anything: you could buy the platform, run the pilot, declare progress, and route around the seam. AI does not sit at that distance. It runs on the authority structure. So the first purchase that actually touches the org design is the first one that exposes it. The trillion dollars is not buying a transformation. It is buying the mirror.

Said, the same day, from inside the building

Easy to say from the outside, easy to dismiss from the inside — so consider who said it, and when. On the very day the survey landed, two people who could not be further inside the building said the same thing in the same publication. The chief operating officer of a 108-year-old, Fortune 500 financial institution and the chief strategy officer of Accenture published a piece with a title that could sit on this site’s masthead: AI won’t fix a broken company. Rewiring it will.5

Their image is a sibling of the watch. Picture a railroad that spends billions on the fastest trains in the world, then runs them on the same aging rails. The trains are not the constraint. The tracks are. Stack agents on top of decades of legacy systems and broken workflows, they write, and you do not accelerate the business — you automate the dysfunction faster.

Point the most powerful tool ever built at an unredesigned process and it does the wrong thing at speed.

That is amplified intention, named by an operator who has watched it happen. Their own words for the conclusion are blunt: AI transformation is not a technology project. It is an operating-model rebuild that happens to run on AI, and companies that treat it as a bolt-on will spend years chasing pilots that never scale.

And because they did the rebuild, they can show the receipt — the upstroke the watch essay said only appears if the down years bought something. After rebuilding the foundation first, cleaning the data, retiring outdated systems, redesigning how work actually flows, plan sponsors can change investment options in days instead of weeks, and digital engagement across millions of participants rose 13%. None of it, they are careful to say, came from a flashy demo. It came from the unglamorous work most companies skip. That is the honest dip, documented from the inside: the boring work is the work.

Read their five focus areas and you find the same three things this site keeps naming: the core and the data, the workflow and the governance rail, the human kept in the loop where trust is the product. But a checklist is not yet a change. The work needs an altitude and a discipline, and the altitude comes first: someone at the top has to own the future state, not delegate it downward. When Block set out to rebuild itself, its CEO did not hand AI to a technology team to go find use cases. He formed a thesis from the top, published it — From Hierarchy to Intelligence — and rebuilt the company toward a single idea: an organization run as intelligence rather than a chain of command.7 Agree with the specifics or not; the altitude is the point. He looked at the future state from the C-suite seat and moved the organization and the technology toward it as one system.

That is exactly what the survey says two-thirds of the C-suite has not done. And it is the shape of the whole argument this site makes: the org and the technology are not two projects but one machine, and the humans rise up to govern it. That frame is HOT — Human judgment, Organization authority, Technology enforcement, held under tension together rather than handed off in sequence. It is the whole board seen at once, which is the altitude the survey shows is missing.

Seeing the board is not the same as moving the pieces, and the Organization dimension is where most rebuilds stall — the analog part that hardened into standard over years of being right, and the hardest to redesign because it looks like competence, not constraint. It rests on a named discipline rather than good intentions: Salim Ismail’s ExO 3.0, the spine for redesigning an organization around intelligence instead of retrofitting it onto one. HOT is the frame; ExO 3.0 grounds the O. The survey proves the seam exists; the operators prove the rebuild pays. This is how you cross from analog to digital: build the second machine, do not bolt onto the first.

What the survey is proof of

Set the finding against the arguments this site has been making, and it stops reading as one more piece of AI news and starts reading as confirmation.

The move the survey hands you

There is a practical instrument buried in the finding, and it is free. The reliable read on your organization’s AI readiness does not come from the level holding the line. It comes from the level below it.

The senior managers said 57% — not because they are optimists, but because they can see the actual decision path: who really signs off, who really gets overruled, where the work really stalls. When you want to know how wide the seam is, you do not ask the person standing on it. You ask the person who has to carry work across it every day. It is the same edge effect the watch essay called the canary: the truth registers first where no one has a stake in the old arrangement.

None of this is an argument that the C-suite is failing and the middle is wise. It is an argument about where the load actually sits. The survey shows leadership teams that each perform well alone and have never had to govern a shared, fast-moving, high-blast-radius capability together. That muscle — governing a capable actor whose decisions you did not personally make — is the same institutional muscle whether the actor is a junior hire or an agent. Most organizations have not built it, because until now they did not have to.

Step back and look at the whole thing

They have to now. The spend already committed to that. What the spend has not yet bought is the one thing it most needs and cannot procure off a shelf: a person, named and load-bearing, who owns the authority envelope and answers for whether the assurance underneath it is real.

This is the realization the whole canon has been circling. It is not a technology decision. It is the moment to re-open the entire strategy — to stop tuning the analog machine and ask whether it is the right machine at all. The $2.5 trillion is the easy part; it is already moving. The hard part is the sentence the survey says two-thirds of the C-suite cannot finish.

The person who owns this is —.

Ask the level below the line. They will finish it for you. And then the real question is not how much more AI to buy, but whether anyone at the top has done what Block’s did: formed a thesis about the future state and moved the whole system toward it — the Human judgment aiming it, the Organization authority that couples it to the work, the Technology that enforces it, designed together as one machine rather than a digital layer bolted onto an analog one. That is HOT, and moving a real firm there is what ExO 3.0 is for. You can see where you stand on all three before you spend the next dollar.

Related on this site: The Back of the Watch · The Problem Was Never the Technology · The Seam · Who Holds the Line · The Loop Moved · The Knowledge Distance Problem. The framework in full: Human · Org · Tech.