Cutler: “Context transfer is not the same thing as shared understanding

Everyone wants AI to fix the mess. The reorgs that never settled. The teams that report to no one clearly. The dependencies nobody owns. Throw agents at it, flatten the org, and let the tools sort out the coordination. That's the pitch a lot of leaders are hearing right now, and some are buying it. Let me catch you up on why it won't work, and what actually will.

The deep cut

• AI amplifies your operating model, it does not replace it. John Cutler's point is flat: a cheap context layer helps a coherent org and speeds a broken one downhill.
• Mandating adoption strips the agency it needs. Cutler shows leaders framing Q3 goals as personal growth, then labeling the pushback a mindset problem.
• You cannot change an org you do not run. Teresa Torres turned down a designer who wanted to fix her PMs, because no sponsor meant no movement.

The fiefdoms you built, now begging AI to save you

The org charts most teams live in were not accidents. During cheap money, leaders gave each area a head, a budget, and a mandate. Autonomy for speed. That created what Cutler calls fiefdoms, and the dependencies piled up until the people closest to the work were left to route around the politics.

Now the same leaders want AI to erase the coordination tax. And it can help. It summarizes, retrieves context, surfaces dependencies earlier. But Cutler draws the line clean: "context transfer is not the same thing as shared understanding." AI does not clarify who decides. It does not resolve contested ownership. It does not create slack.

The trap is coordination theater. People who already avoid working with others get a new set of tools to keep avoiding it. Their agents talk to your agents. Async work piles up faster than anyone can absorb. The old coordination burden got pushed to the front lines by org design. The new one gets pushed there by async abundance.

The org chart is a funhouse mirror, not a window

Before you point AI at "the truth" of how your org works, know that there is no single truth to point at. Cutler's take on mirrors is that organizations are multi-party. Dozens of people hold up fragments, each angled differently, each with something at stake in what gets seen.

That is why defining a team is so hard. Cutler lays out the frames: product sees capabilities, design sees end-to-end journeys, engineering sees services and code ownership. A designer sees one continuous experience where product carved out three areas and engineering split five services. The skill to name this is low. The political cost is high.

So people frost the mirror. New team names hide old strategies. And Chris Argyris named these gaps "undiscussables," things everyone sees but no one is safe to say. AI can render the map. It cannot make the room safe enough to read it out loud.

Mandates kill the thing that makes adoption work

Here is where leaders keep tripping. They treat resistance to AI as a personal failing and frame adoption as a motivation problem. Cutler catches the sleight of hand: the language says "this is for your growth," the structure says "this is for our Q3 goals." One person's rant sums it up: "Give me a real problem, give me space to figure out whether AI helps solve it, and get out of my way."

The math backs the skepticism. Cutler asked a CEO who claimed to be 100x more effective whether there was 100x more demand for their product. Crickets. Capability grows, but the judgment about where to aim it does not.

The SCARF model names why mandates backfire. Status, certainty, autonomy, relatedness, fairness. A threatened brain gets less creative, not more. Tell a twenty-year veteran their worth is now measured by prompting, and you have triggered a threat response, then blamed them for it.

Change starts with you, and only you

If you cannot mandate your way to adoption, and AI cannot fix your structure, what moves? Less than you want. Teresa Torres got an email from a designer who wanted help training her PMs. She said no. Without executive sponsorship, felt pain, and urgency, nothing moves.

Petra Wille's framework is three conditions: pain that leadership feels, a real cost to inaction, and awareness that solutions exist. Miss one and you are pushing a boulder alone. Torres puts it plainly: "Organizations don't want to change. What they're doing today is exactly what got them where they are."

So work within the constraints. You do not need to rewrite quarterly planning to do good discovery. Show your work instead of arguing your conclusions. Teams that do great work in the open become the ones others copy. That ripple spreads further than any mandate.

Three questions for your team

• Before we point AI at coordination, can we name who decides, who owns the end-to-end experience, and where those two do not match? If we cannot, AI will only move the confusion faster.
• Can I say out loud who specifically benefits from our AI push, and over what time horizon? If the honest answer is "our Q3 metrics," the team already knows, and the growth framing costs me trust.
• Where are we trying to force change without sponsorship or felt pain, and what would it look like to show our work instead and let one good team become the example?