Find Your Buddy: The Billion-Dollar Company Starts With Two People

In early 2024, Sam Altman said his tech CEO friends had a betting pool.

In my little group chat with my tech CEO friends there’s this betting pool for the first year there is a one-person billion-dollar company, which would’ve been unimaginable without AI. And now [it] will happen.

— Sam Altman, early 2024 (via Every)

A year later, Dario Amodei was asked when. He said “2026.”

It is late 2026. The closest case is Medvi: $401M in 2025 sales, one founder, and his brother. No valuation. An FDA warning letter. Even the NYT said it is “technically not a one-person $1 billion company.”

I think the bet has the wrong unit. It is not one person. It is two.

Professional Problem Solvers

Developers were always professional problem solvers. For thirty years, most of that was the doing: typing, debugging, wiring things together.

Agents took most of the doing. I wrote about that in the last post: the craft moved up a level.

What is left is the “professional” part. Judgment. Knowing what to build, what to cut, and when to say no.

And you can still be all five archetypes. One dev with a good harness covers the whole build now.

But the whole build is only half a company.

Judgment Needs a Problem

Judgment does not run on nothing. It needs a real problem and a real customer. You do not find either in a terminal.

• Agents can build. They cannot get a meeting.
• Agents can write the pitch. They cannot be trusted by the buyer.
• Agents can research a market. They have not spent twenty years in it.

If you have the product and not the channel, you don’t have a business yet. You have a demo.

— Dustin Deno, Oct 2026

So find your buddy. Someone who sells, or someone who knows a domain cold and has a network in it. Then be their partner.

FameCake

FameCake puts anyone on a billboard in a couple of minutes. Mike and I run it as partners.

We met through a personal intro. The split is simple:

• Mike does the deals. He works with the partners and the media owners. He has spent years in out-of-home advertising.
• I make every technical and product call. All of them.

A month in, I told Mike the honest version: “I can tech the shit out of anything but I cant OOH the shit out of anything yet.” That is the whole post in one line.

Push and Pull

A buddy is not a client. A client pulls. A partner pushes back. It is push and pull, both ways.

• I cut a feature a big media owner liked. We had tested AI style transforms. My summary: “nobody cares.” The data backed it, so they went.
• I kept credit expiry at 90 days. Mike wanted it shorter. I thought a short deadline would make customers feel herded. I sent him the repeat-purchase data, and 90 days shipped.
• I reshaped his pricing flyer. I kept his goal, longer bookings, and dropped the price rise on one-day posts. Those are the trial product.
• He held back my house-ad test. I wanted a price on every house ad. Mike knew one partner would not like it. That is his judgment, not mine, and he was right to use it.
• He gave us the line. Mike said “billboards on demand.” It was on the homepage the next day.

It Is Not Just FameCake

I see the same split in other ventures I can’t name. A domain expert in a regulated field with a problem nobody solved. A client who owns the customers. A dev who can build anything. Each time, the pair works and the solo version stalls.

The two-person exits are real too. FlashDocs was an ML dropout and a veteran entrepreneur who met at a VC dinner. They sold to Hebbia in 2025 (Fortune).

The Odds

In the persistence post I said persistence buys attempts, but the odds per attempt stay flat.

A buddy is how you move the odds per attempt. You build fewer things nobody wants. You get in front of buyers on day one.

Paul Graham made “single founder” mistake number one in 2006. Agents did not change that. They made it more true. The build side got cheap. The other side did not.

Find Your Buddy

This is the stance: every dev should do this now.

• Be all five. Own the build end to end.
• Find someone with a problem and a network. Not someone with an idea. Ideas are free.
• Use personal intros. Mine came from one.
• Be a partner, not a contractor. Equity, not hourly. Push back.

The billion-dollar company might come. When it does, I bet it has two names on it.