One Dumb AI Question Worth a Fortune

From an email I got today.

Why do apples fall?

Sounds like a dumb question.

But when Newton asked it, he discovered the same force that keeps the Moon in orbit.

Why is the sky dark at night?

Another dumb one, indeed.

But that question cracked open the modern universe—and led to the theory of the Big Bang.

Why does bread rise?

Who cares!?

Louis Pasteur did. And in asking, he stumbled into modern medicine.

Playing stupid games might win you stupid prizes…

But asking stupid questions built everything we know—and might just help us see what comes next before anyone else.

Why Do Companies Exist? 

Back in 1937, economist Ronald Coase asked a similarly stupid question: Why do companies exist?  

His answer: Companies exist because doing business is expensive.

You don’t start a company because you love org charts and status meetings. You start one because working with people—finding them, hiring them, paying them, making sure they do what they promised—is a logistical nightmare.

So we built hierarchies. Put everyone under one roof. Called it a “firm.”

As Coase saw it, companies are just machines for making transactions cheaper.

BUT, here’s the rub…

What happens when transactions themselves become virtually negligible? What happens when the cost of coordination drops to nil?

Do we even need companies?

According to a group at MIT, perhaps not. 

The Coasean Singularity

Their paper, published by the National Bureau of Economic Research, carries a title straight out of sci-fi: “The Coasean Singularity.”

Premise? The next trillion-dollar market might not come from a new industry…

But from the collapse of transaction costs.

“The fundamental economic promise of AI agents,” they wrote, “lies in their ability to dramatically reduce transaction costs—the expenses associated with using markets to coordinate economic activity.”

AI developers envision a world where AI agents act as our proxies in commerce, seamlessly negotiating, transacting, and contracting in milliseconds.

If such a future materializes, machine-to-machine coordination would compress transaction costs to virtual dust, reshaping the logic of the firm.

That idea raises two questions worth asking: What does that world look like? And what does that mean for those trying to get ahead of the curve?

How It Starts: The Demand Side

As you read this, AI agents are being built to find, negotiate, and execute.

They don’t sleep. They don’t need health insurance. They don’t “circle back next week.”

They just do.

Of course, nobody wants an AI agent just to show off. They want it because it kills the wait between wanting something and getting it.

The real challenge in any market isn’t always production—it’s allocation. Creating goods is hard, but matching them precisely to individual preferences is even harder.

Every person has unique preferences, contexts, and constraints.

Traditional systems—retail, finance, healthcare—struggle because personalization doesn’t scale well. It’s expensive to understand, predict, and deliver exactly what each individual on the planet needs and wants.

Enter AI agents:

• “Find me the best house in Austin with a backyard and solar panels.”
• “Negotiate my freelance contracts under $5k.”
• “Rebalance my portfolio every Monday, focusing on the top revenue-earners in energy, silver mining, and AI.”

Nobody wants to scroll through infinite listings or fine-print. They want outcomes. People will adopt AI agents when the effort saved beats the effort spent. 

Imagine telling your AI assistant, “Find me a developer who can build this feature for my app for $3,000.”

A minute later, your AI has scanned 10,000 freelancers around the world, vetted them, negotiated price, verified credentials, and sends you both the contract to look over.

Multiply that by millions of interactions, and the economy stops being a bunch of people emailing PDFs and becomes a swarm of digital agents constantly transacting, 24/7.

And if you zoom out, this is what every big technological revolution does: it kills a cost. 

Steam power flattened the cost of distance. The internet killed the cost of communication. AI agents could do the same for the cost of coordination.

Early adopters? Anyone facing too many choices and not enough time—job seekers, freelancers, online shoppers, investors drowning in options.

In other words, pretty much everyone. 

The Supply Side: New Gold Rush

On the other side, firms are already lining up to build, integrate, and monetize these agents.

Some will build closed ecosystems, like Apple-style walled gardens where their agents only talk to each other.

Others will build open protocols, cross-platform agents that roam freely like economic nomads.

Either way, the agent economy will spawn its own design wars:

• Who owns the agent?
• Who sets the rules of negotiation?
• And who gets the house cut?

It’s the same platform dynamic we saw with App Stores, ad networks, and crypto exchanges. Only now it’s about who controls your agent.

Our longstanding thesis: a swarm of small, specialized AI models will outperform the giant, monolithic ones Big Tech is obsessed with.

If we’re right, then the irony is thick—in their race to build “God Models,” the tech giants are actually paving the road for their own disruption.

Every new model, every open-sourced tool, every layer of infrastructure they release makes it easier for anyone to make smaller, faster, more focused models—often without even knowing how to code.

In plain English: They’re not building AGI. They’re building the tools that will eat their breakfast, lunch, and dinner.

The Big Picture: Markets Rewritten

Here’s where it gets even more interesting.

If agents handle preference discovery, contract enforcement, and identity verification—the three hardest problems in economics—then the entire architecture of the market changes.

We could see:

• Continuous micro-contracts replacing static employment.
• AI-to-AI auctions running every millisecond.
• Decentralized negotiation layers where trust is encoded, not assumed.

Even if the guys at MIT are right, this won’t happen overnight.

But if it’s even close to being directionally correct, the time to position for the windfalls along the way is now.

What we do know:

Every time humanity has lowered a fundamental cost, the world reorganizes.

It happened with the printing press.

It happened with the steam engine.

It happened with electricity.

It happened with the internet.

And now, with AI, it’s happening again: maybe even with the cost of coordination itself.