A guy in LA launched a telehealth company from his apartment eighteen months ago. He spent $20,000 and used a stack of AI tools to build the whole thing. No investors. No team. The New York Times reported that his company, Medvi, did $401 million in sales in its first year and is tracking toward $1.8 billion in 2026. His total headcount is two. Him and his brother. Sam Altman said it looks like he won the bet.

This story is everywhere right now and everyone is telling the inspirational version. Let me tell you the version that actually matters if you're building something.

He didn't build an AI product. He used AI to build a distribution machine for an existing product. GLP-1 weight loss drugs. The doctors, the prescriptions, the pharmacy, the shipping — all outsourced to partners. What he built was the front door. The website, the ads, the checkout, the customer service. All powered by AI. All run by one person. The insight here isn't about AI creating something new. It's about AI making something that already exists dramatically more efficient to deliver. That distinction matters because most people are still chasing the "build a new AI product" dream when the real money is in using AI to do existing things better, faster, and cheaper than everyone else.

Now here's the part that's getting less attention. Six weeks before the New York Times ran that glowing profile, the FDA had already sent Medvi a warning letter for misleading consumers about its products. Since then, investigators have found over 5,000 ads running on Meta under fake doctor personas — AI-generated profiles promoting the company's drugs. A class action lawsuit was filed in March. Deepfaked before-and-after patient photos. A Forrester analyst published a piece titled "Beware The Magical Two-Person, $1 Billion AI-Driven Startup."

When you're the only human reviewing AI output, the error surface is enormous. I know this firsthand. I accidentally spent $2,000 in a week on API costs because my prompt was too bloated and I was running it without thinking about the cost per query. Different scale, same lesson. AI moves fast. If you're not paying attention, it will move fast in the wrong direction.

The real takeaway from the Medvi story isn't "one person can build a billion dollar company." It's that the distance between idea and execution has collapsed so completely that the thing that matters most now is knowing what to build in the first place — and having the judgment to know where the lines are. AI handles the execution. But taste, judgment, timing, ethics, and knowing which game you're even playing? That's still you. That's always been you.

  • If the Medvi story has you thinking about your own idea, go test it before you build it. I made BigMoneyIdeas for exactly this reason. It scores your idea in two minutes and tells you where the gaps are.
  • Next week I'm writing about the AI company that just got blacklisted by the US military. Not for building bad technology. For refusing to remove its safety guardrails.

🎵 Giorgio Moroder — Chase 1978. The original soundtrack to moving faster than anyone thinks is possible. Still sounds like the future.

— Nicc