What You’re Actually Buying
When You Buy AI.

I was traveling most of this past week, and there was one thought I could not put down the whole time. It started small, almost a side observation, and by the end of the trip it had become the thing I most wanted to say to you. So this is a field note. Shorter than usual, a little looser. Me thinking out loud about something I think is getting missed almost everywhere right now.

Here it is. Almost every organization on the planet is buying AI right now. And almost every one of them believes it is buying the same thing. Efficiency. A faster, cheaper way to get done what people are already doing. That is the story the technology tells about itself, and it is the story leaders are repeating back in their board meetings and their all-associates meetings. I think it is the wrong story. And the gap between that story and what is actually happening is going to decide which organizations are any good a few years from now.

Because here is what you are actually buying, underneath the whole productivity case. You are buying a change in the kind of thinking your organization runs on. Every time you bring AI into how the work gets done, you are making a quiet decision about whether the people in your organization get sharper, or slowly stop thinking at all. And almost nobody is treating that as the decision. It never makes it onto the agenda.

So let me sit with why it stays off the agenda, because I do not think this is carelessness. I think it is inherited.

Every organization is running, underneath everything, on an operating model that was built for a factory. And a factory has a very particular attitude toward thinking. It treats thinking the way it treats a machine. Something that should run automatically, at the lowest possible effort, that you measure by how much it produces and how cheaply. In that world, cognition is not something you design. It is just assumed. It is like the lights coming on. The only interesting question you are allowed to ask about it is how to make it cheaper.

Now watch what happens when AI walks into that assumption. AI looks like the dream come true. It gets sold as automation and efficiency, the factory's oldest wish finally reaching all the way into thought itself. So leaders receive it exactly the way the factory trained them to. As a cheaper way to produce the same execution.

And that reception is the whole mistake. Because the real event is not that execution got cheaper. The real event is that execution is leaving. And what is left, what actually decides now whether your organization is any good, is the high-effort thinking the machine cannot do for you. So when you treat a cognitive event like a technology event, you quietly optimize away the one capacity that was about to matter most. One efficient little task at a time, while everyone in the room congratulates themselves on the savings.

And the strange thing is, the culture is already half-noticing this. There was an essay in the Atlantic recently that put the human version of it really plainly. It argued that in the age of AI, what is going to separate people is not how smart they are. It is their relationship to mental effort. Some people keep thinking hard, and they expand what they are capable of. Other people let the machine think for them, take whatever it hands back, and slowly lose the ability to push back on any of it.

I think that essay is exactly right about the split. But look at where it goes for the fix. Personal discipline. Start from a blank page before you go to the bot. Ask it for hints, not answers. Don't let it write the things that actually matter. And that is good advice. For a person.

It does nothing for an organization. Because an organization does not have willpower. It has a design. You cannot fix this one person at a time, by asking everyone to be a little more disciplined. The thing that decides whether your organization keeps thinking is not the willpower of the people inside it. It is the architecture around them.

And I will be honest with you, I see this from the inside in a way I cannot really turn off. My own mind runs the long way around by default. I do almost everything through deliberate effort, with very little of the fast, automatic intuition most people get to lean on. So the line between thinking that actually costs you something, and thinking that has quietly been handed off, that line is not abstract to me. I live on it. And it turns out to be exactly the line the factory model was built to pretend does not exist.

So if you are a leader, the real question is not whether your people have the discipline to keep thinking. It is whether your organization is built to make them think. Whether it runs on high cognition or low cognition is not luck, and it is not the sum of everyone's good intentions. It is an architecture. You either design it on purpose, or you inherit one by default. And it is built out of completely ordinary things. Which roles you expect to actually exercise judgment, and which ones you expect to just execute. Which skills you develop in people, and which ones you quietly automate away. What people actually get rewarded and promoted for. That is the real work of leadership now. Designing how thinking happens in your organization, so that high-effort cognition is the actual job people are there to do, and the reason they stay.

And if that sounds like some big new demand I am putting on you, I want to point at something you already do. At the very top of the organization, you already accept exactly this principle. A board does not leave it to chance whether the information that matters reaches the people who decide. It builds reporting lines. It builds the board pack. It builds an audit committee, a confidential channel. And it does all of that because the law treats getting critical signal to the top as a duty, not a hope. The names that duty travels under, if you have ever sat near a board, are Caremark and Sarbanes-Oxley. And the principle underneath them is the exact thing I have been describing. You cannot assume the signal just arrives. You design the route it travels. And the responsibility for designing that route sits at the top. And that duty keeps widening. It started with financial signal. Then safety. Then cybersecurity. Now it is reaching AI. Every time a new category shows up that can blindside the enterprise, it gets pulled onto the board's standing job of making sure it can see.

But there is one signal that has never been pulled onto that list. I have started calling it the cognitive signal. It is the judgment that somebody in your organization has actually done the work to reach, on its way to the decision that needs it. Picture someone who has reasoned a hard call all the way through, and gotten it right. They can see why a plan is not going to hold, or why a number everybody is treating as solid is sitting on an assumption that does not. The judgment is sound. And it dies before it ever arrives. Because the organization has a whole system for moving reports and numbers, and nothing built for moving a judgment. So the decision gets made without the one piece of thinking that would have changed it.

And here is the part that took me a while to really see. The judgment that is most worth having is very often the judgment that travels the worst. Because the organization rewards the version that shows up fluent and certain, and it has no way to receive the one that shows up careful, and qualified, and slow. Nobody owns that channel. No reporting line is required to carry it. It is the exact same failure the board already governs against everywhere else, the right information never reaching the people who decide. The only difference is that here the signal is a judgment. And a judgment is the one kind of signal nobody was ever assigned to carry.

Now I want to take a turn here, because this is also, I think, the answer to the fear sitting underneath all of this. The fear that AI just means the work goes away. That the jobs disappear.

That fear has the exact same shape as the efficiency story. It assumes a job is a quantity of doing. So once the doing gets cheap, the person looks redundant. But the doing was never where the value lived. In the brain economy, the returns come from cognition, from the quality of judgment a system can actually bring to bear. And good judgment does not just protect a job. It makes all the execution underneath it better, and more thorough, because an organization that thinks deliberately makes fewer wrong turns and catches more of what matters, everywhere the work runs.

So an organization that really gets this does not spend the moment defending headcount. It does what the last great displacement did. When the industrial revolution made physical production cheap, the work did not disappear. It moved up a level. Into whole industries and organizations and products nobody inside the old frame could have pictured. And the economy grew past anything that frame could even measure. I think the same room is opening up right now, one level up, in cognition.

And that is the question I actually think is worth putting on the agenda. Not how many jobs AI is going to remove. But what a cognitively charged organization would build, once execution is cheap and fast, and the scarce, deliberately designed resource is how well the place thinks. We do not have the answer to that yet. We have never built one on purpose.

And building one on purpose starts smaller than that big question. It starts with designing the path that judgment travels, the same way a board long ago learned to design the path its critical signal travels. The first move there is the one most leaders keep misreading as a communication problem. The read that did not land, the point that did not carry. We file it under communication. But communication is really a function of judgment. It is the system that carries a judgment from the person who reached it to the place it is needed. And when that system is undesigned, the judgment gets lost. Designing that system is what the next arc of this work is about. It is called the Cognitive Translation Protocol, and it is the first of three moves. Gather the signal. Then settle who actually holds the judgment on it. Then route the work, so judgment can keep up with the pace. Three names for one act, really. The deliberate design of how an organization thinks.

So let me leave you with the thing I could not put down all week. You already govern for the signal that has to travel. You have just never had to design the cognition that produces it. So this week, watch for it. Watch for the moment when something was obvious to the person who saw it, and it never made it to the decision it belonged to. That is the signal with no system. And that is where we start next week.