Intelligence Per Dollar: The New Economics of Business

When you plan your budget next year, you won’t plan around headcount anymore.

You’ll plan around the intelligence each team needs. Every team lead will get a budget, and they’ll decide the mix of human intelligence and artificial intelligence (AI) that gets them to the goal. A friend at a public company told me this is what they’ve already started doing.

We used to be reliant on humans for intelligence, but AI has advanced to the point where it can actually replace many human reasoning tasks. We’re moving away from human resources to the unit of the future: Intelligence per Dollar.

How we plan will totally change. What work needs doing, what work do we want our company to do, and how much are we willing to pay for it? That question is as old as business itself. What’s new is that, for the first time, we can actually answer it with real numbers. And importantly, measure and optimize.

Why Intelligence per Dollar, and Not Tokens

You’ve probably noticed the industry quietly backing away from the word “token.” Companies are swapping in “credits” and other repackaged units, and there’s a reason for it. Different models spend tokens differently, so a token here doesn’t mean a token there. It was never a unit you could compare across the board.

But tokens were always beside the point. The real question, the one every business has been circling forever, is simpler: what does it cost you to deliver the goods, services, and products you put in front of your customers? That’s the number that matters. Intelligence per Dollar is just the honest name for it in the age of AI.

The dream we could never reach with humans

Here’s what makes this moment interesting. In a way, this is the thing we always wanted from human resources and never got, because humans are just too dynamic to pin down.

The dream went like this: You’d measure the exact output of every employee. You’d take what each person produced on every project, spread the cost across those projects, and see clearly what time was creating value and where it was leaking away. Who was doing better, who was doing worse, who was costing more than they returned.

It was never possible. And it was never possible for reasons that stacked on top of each other.

Measurement itself was the first wall. It’s genuinely hard to know what someone does all day, and the moment you try to track it too closely, people push back. They feel micromanaged, watched, reduced to a number. So real measurement mostly never happened.

Then there’s the simple fact that people are people. We have good days and bad days. We get sick. We show up motivated or we don’t. There’s an entire industry devoted to keeping employees engaged and productive, which tells you everything about how uncontrollable that variable is.

Even if you could measure output, output refused to sit still. It shifted from person to person. Teams had chemistry, working better or worse depending on who was in the room. People moved between projects week to week, picked up new work, crossed into other departments. By the time you accounted for all of it, cleanly measuring anyone’s contribution was effectively impossible.

And that was only half the problem. The other half was that even when you saw something worth changing, you couldn’t just change it. You couldn’t tell someone to stop and change what they work on day to day, or increase or decrease a team’s time on a project on a whim. There’s office politics. Hiring and training takes time. Firing is a big deal. You couldn’t add and remove parts of someone’s day, or parts of their energy, on demand.

So business lived in broad buckets. You grouped teams together and did your best to approximate the cost of goods sold, approximate what projects cost, approximate what people cost. Whether an employee was net positive or net negative came down to broad strokes, usually judged quarterly against whatever performance metrics a manager assembled, never really reaching the nitty gritty of the finances.

And now we can

That’s what’s changed. It’s a genuinely different world.

Once you start using AI to do productive work, in more or less the way a person would, you can finally do the thing that was always out of reach. You can measure, at a micro level, exactly what’s being done and exactly what it costs. Broken out by project. Broken out by task. You can get into the weeds of where the money is actually going, and then move it. Cut what isn’t productive, increase what is. Reflect the true cost of the services, products, and projects you’d otherwise only be guessing at.

Think about how hard this used to be. If Sally spent an hour a week on a project, good luck folding that cleanly into your financials. But AI spend leaves a trail. Take your token or cloud spend for that usage, and you can finally see what it costs.

The best part is that scaling is simple. No hurt feelings, no burnt out teams that need a vacation. No HR policies, no onboarding time. You increase what’s working and cut what isn’t, without anyone shedding a tear.

That pulls companies into an entirely different way of thinking about cost and planning. And because the measurement is suddenly this precise, I think it introduces a new responsibility at the center of every finance and leadership role: knowing exactly where your AI is deployed, and exactly which people and teams it’s serving.

Right now, almost nobody is doing this. Which is precisely why the opportunity is so large.

There’s a real chance to understand, and then continuously adjust, the Intelligence per Dollar a company spends, something that was effectively impossible when your intelligence came from humans. It opens the door to optimizing, planning, and having honest conversations about the questions that were always fuzzy. What are we paying? What does it cost us to build a new project, to grow, to deliver our services, to experiment? And to answer all of it with an accuracy we’ve simply never had access to before.

This only grows from here

And it compounds.

Every company is going to keep increasing the amount of AI it uses, which means this question gets bigger over time, not smaller. For businesses that end up spending hundreds of thousands, or millions, on AI, being able to explain where that money goes, both to themselves and to their stakeholders, stops being a nice to have.

Because AI is dangerously easy to run badly. It’s easy to pour spend into low ROI and even negative ROI projects without noticing. I call it “ease inflation,” where building becomes so frictionless that you end up building because it’s quick and simple, but it contributes nothing to your revenue while quietly costing you. It adds up fast.

So this is the shift. The economics of business are being rewritten, and Intelligence per Dollar is the unit we’ll be measuring in. The companies that catch this early, that actually get into the weeds of where their intelligence spend goes and why, are going to understand their own business with a clarity that wasn’t available to anyone before.

And there’s a new gap in the market waiting to be filled. Every company, from the smallest startup to the largest enterprise, is going to need a way to track, manage, and plan for their AI spend. We’re talking about anywhere from 10% to 80% of a company’s current human resources budget going toward AI, likely spread across many different service providers and products. Managing that is going to be critical, and right now almost nobody is built to do it.

We’re moving away from measuring cost and revenue by employee. The new denominator is “Intelligence.”