The empty investor's chair

You have a new job.

You didn't apply for it. Nobody announced it. There was no offer letter, no onboarding, no training.

It started the day someone handed you the tokens.

And you have been doing this job for months now, whether you knew it or not.

Right now, this new job is being given to people across organizations big and small.

In December, Microsoft handed Claude Code to thousands of its employees. Not just engineers. Project managers. Designers. People who had never built software were invited to build.

Usage boomed.

Six months later, Microsoft cancelled most of those licenses. But the tool was not pulled because people disliked it. It was pulled because they used it too much.

Amazon ran the same arc. It built KiroRank, an internal leaderboard that ranked employees by their AI usage. A scoreboard for spending. Then it tore the leaderboard down.

Both companies could see everything they were spending. Neither could see what it was returning.

Two companies. One motion.

Unlimited permission to spend. Then severe restriction.

Eat as much as you want. Now I'm going to starve you.

Feasting, starving. Neither builds a healthy body.

Unlimited tokens, token starvation. Neither builds a more valuable you.

Neither extreme is deliberate. Neither one requires discernment. Neither looks like investing.

Yet investing is the only way to move forward. 

Look closely at what those organizations permissioned. Tokens. Only tokens.

But tokens never travel alone. Every token comes at a cost. And I don’t just mean their literal, monetary cost but also the cost of the time you spend. Your energy. Attention. Imagination. Nobody wrote a memo granting permission to spend imagination. It rode along, unnamed.

We never had to think about spending those things before. In yesterday’s execution economy, those budgets were spent for us. 

The decisions were made before the work ever reached us. The steps to execute the decisions were clear. We didn’t have to ask what any of it cost. It was already spoken for.

Now, we have to decide for ourselves what is worth spending.

Yesterday’s execution economy was a Can economy. Can we build it? Can we run it? Can we hit the date?

Can was scarce. So a small number of people were paid to hold Can and Should together. Think of a product leader. An endless feature list. Limited capacity to build. Someone had to ask both questions at once, and it was their job title. They decided what should move ahead.

That mindset is not new to the world. It was a job. But now it is everyone's job description.

AI did not make all of Can free. It expanded what is possible and made more of it free. And the window keeps expanding as AI evolves. Three years ago, you could not vibe-code an app you would actually use. Today you can.

I know because I’ve done it. I have poured time, energy, attention, and imagination into vibe-coding things I never asked “Should" about. Just because I can does not mean I should.

What is left standing at every desk is Should. 

“Should I?” is the question that the playbook never asked you, because someone above you had already answered it before the work arrived.

That is where the cost went. The cost of Can collapsed. The cost of Should went up. Because Should still runs through the only things that are finite: your time and your attention, and the time and attention of whoever receives what you build.

The constraint was never tokens. Tokens are replenishable. They cost money, but you can always buy more.

Imagination also does not run out. Judgment does not run out.

However, your time and your attention are the only things here that are actually finite. And they are the things nobody told you that you were spending.

Imagination and judgment can only be applied through time and attention. Which makes those two the binding constraint on everything else.

You already hold a portfolio of AI bets, whether you have called it that or not. What are you investing your tokens, your time, your attention, your imagination in? What is coming back? 

Some of what you build stays local to you. Some touches your team. Some reaches the organization. Either way, it compounds in both directions. Invest well, and the returns compound positively, even when the thing you built never leaves your desk. Invest badly — because you can, without knowing whether you should – and the opportunity cost compounds negatively instead. 

The waste was never the tokens. The waste was what you did not make instead. 

Why? Because most of what people build never travels upward. What reaches the top of the organization is a bill and no visible gain. Spend went up. Nothing moves faster. No new possibility appears.

The top is not looking at the wrong number. They’re looking at the only number they can see.

ROI gets measured against token spend. But time, attention, imagination do not travel and do not get measured. The root cause cannot be seen, so it cannot be attacked. The only lever the top can see is spend. So they cut spend.

And yet, restricting token spend is not the answer either. Poorly investing your imagination is one mistake. Asphyxiating it is worse.

Throttle the tokens and you throttle all of it by default. Nobody sent a memo cutting imagination. They did not have to.

The thing nobody named on the way in is the thing nobody defends on the way out.

So the reversal is not a swing from too much to too little. It is a swing from wasting the unnamed things to eliminating them. But because the second one looks like discipline, nobody stops it.

Neither extreme is the answer, because both answer the wrong question. The right question is not how much. It is toward what. And answering it means each of us sitting in the investor's chair.

Yesterday's technology investments were a single person's call. Someone decided whether the ERP would pay for itself. They signed. Everyone else received the decision and used the system.

One chair. One occupant. And that was correct. You did not need ten thousand people deciding whether the ERP would generate a return.

AI is different. The possibility of what one person can make has exploded. The ability to act on that possibility arrived at every desk, in every function, at the same time. And both keep expanding.

Everyone wields AI directly. So the investment decision distributed itself the moment the tools did.

Nobody moved the chair. It multiplied.

At the same time, AI is getting easier for us to use every month. Because AI consumes its own complexity, its capabilities are constantly evolving. The span of possibilities is widening for each of us.

And as the tools expand our ability to execute — the Can space — it gets harder and harder to decide on whether we should.

That’s why occupying the investor’s chair is now non-negotiable.

Every person holding tokens and tools is now making investment decisions. What to build. What to keep. What to kill. What deserves their attention.

That is not a perk. It is not an opportunity.

It is a job. And it comes with accountability to produce return.

The failure I keep seeing in companies across industries and of all shapes and sizes is not personal or individual. It is distributed. You didn't know the chair was yours. It was never named. Never acknowledged. Never trained for.

Your team leader didn't realize they now lead a team of investors. The organization was never designed for any of this. Each level's silence licensed the one below.

Some organizations will object here. “We did name it,” they'll say. We have a Chief AI Officer. We made the CFO accountable for AI ROI.

But they only put one person in one chair. And that person holds accountability for a return that only thousands of unoccupied chairs can produce. One investor, alone, in an organization that does not know it is supposed to be full of them.

The chair is not a seat. The job is not a title.

It is two opposed temperaments, held at once: Build fast and build many, and expect most of it to die.

Move like an entrepreneur. Allocate like an investor.

The entrepreneur in you wants to build everything. The investor in you expects most of it to die. Hold both.

A seat can be assigned to someone. A temperament cannot.

The temperament we must command from now on is one where we hold Can and Should together, at once.

That is why the new job feels foreign. It is not a harder version of your old job. It is the other half of a question you were never handed.

So what does Should sound like?

Am I solving a real problem? For whom? Is it in their top three?

Will this still be a problem in twelve months, or will something else solve it, or dissolve it, before I finish?

If I stood a year out and looked back, would I still build this?

And the hardest one. Of everything they could consume, built by others or by themselves, why would they choose mine?

Should filters for value, not permanence. Building something disposable is a legitimate answer, when you choose it deliberately. What Should forbids is not short-lived work. It is unexamined work.

Even an internal buyer is buying. They pay with their energy and attention.

And every “Should” question ends in a decision. Keep. Kill. Feed. Including the investor's hardest move: killing things that work but do not matter. Some of those decisions will be wrong. That is how the judgment gets built.

Where these questions live in your workflow — before the prototype, before the customer, before the ask — is yours to decide.

That is the conversation I'm taking to my LinkedIn Live next week. Details below.

The job is yours. It has been since the day they handed you the tokens. But it was never just the tokens. The capabilities keep getting easier. The possibility space keeps getting wider. The job grows every time the window does.

Nobody told you. Now you know.

So the question left is not whether you spend within budget. 

The question is whether you invest with an eye toward return. Return on your time. Your energy. Your attention. Return on the possibility space everyone now stands at the edge of, with equal access.

If everyone holds this job, leaders have to lead differently, and organizations have to operate differently. Yesterday, a handful of people could inhabit the possibility space. Now access has been democratized, and the ability to act on it came with it. 

New possibility can come from anywhere on the org chart. That does not match the  white-collar assembly line that most companies are running today. It is something closer to a living organism. Or a startup accelerator.

That is where we go in the next edition.

Right now, I invite you to sit with four questions. Two for you. Two for whoever you lead.

That's what I'm watching for Tomorrow.

– Nish

On next week's LinkedIn Live (August 19, 12pm ET), three ideas I left out of this edition: 

A gratitude P.S.

A few weeks ago, I visited my aunt in London. She is 102.

She was born in Kampala, Uganda, raised seven children, and has lived in London for decades now. Two years ago we all gathered there for her 100th birthday. She had already survived COVID. A short while after, she fell and broke her hip. For a 60-year-old, that can be the end. Let alone at 100.

She had surgery. She went home. She was bedridden. She marched forward anyway. She turned 101. Then she started getting up again. Sitting in the sofa chair beside her bed. Occasionally walking to the loo, as they say in England.

A couple of weeks after this photo was taken, she turned 102.

She was never formally educated. The rules of her generation, her culture, and her gender wrote most of her destiny for her. She raised a family through all of it and came out one of the wisest people I know. 

Her wisdom was earned lesson by lesson. What worked. What didn't. Her wisdom isn't her words. It's her example. One that has greatly influenced me. It shaped her children. It's shaping their children. And it will keep rippling, through people she will never meet.

Nobody ever asked her to be remarkable. She was and is anyway.