Day 14: $5M in 90 Days With AI — What I Learned

Eight years running physical product businesses. I know what it takes to scale from zero. But right now you can build an entirely new business from scratch with AI doing most of the heavy lifting. I wanted to test exactly how true that is. So I challenged myself: start completely over, enter a brand new category, build with AI — no large dev team, no months of runway before launch.

Why Software?

Physical product businesses have real constraints: inventory, logistics, shipping, returns. Software has none of those. When you can build software with AI as a non-developer, the economics become extraordinary. The margins are different. The speed is different. The scalability is different.

You cannot rely on AI to make decisions or build anything on its own. You need someone to delegate, direct, and push it forward. This was the most important thing confirmed in the first two weeks.

What Decision Making Looks Like Day-to-Day

• Every morning: What's the most important thing to build today?
• Review and course correct: What did AI build yesterday? What's off?
• Judgment calls: Product direction decisions AI can't make on its own
• When "good enough" is good enough: Move forward, don't polish forever
• Ruthless prioritization: AI can do a hundred things — you pick the ten that matter

Two years ago, it would take a developer a week just to edit something on a website. In this build: a full website with AI functionality in one week. With a $200 subscription. As a non-developer. That sentence deserves to sit for a second.

Day 14 is launch day. The product is live. And now the real work begins — the same work that's been done with physical businesses for eight years: getting customers, testing what works, scaling what does.

The difference this time: all the AI systems built for previous businesses plug directly into this new one on day one. The media buying agents. The customer service agents. The analytics. The creative pipeline. Systems that took years to build, running from launch day.

Every agent system you build becomes an asset you own. When you start a new project, you're not starting from zero. You're plugging in proven systems and adapting them. Your tenth business launch will be dramatically faster and cheaper than your first.

You don't need 90 days. You don't need a dev team. You need a clear problem to solve, a $200 subscription, and the willingness to make decisions fast.

• Pick the category — what problem are you solving? For who?
• Get Claude Code — $20-200/month depending on the plan
• Define your daily decision rhythm — what's the most important thing to build today?
• Build the MVP — minimum viable product in plain English, iterating daily
• Set a launch date — pressure creates focus. 14 days is enough.
• Plug in your existing AI systems — don't rebuild what you already have
• Launch and start the real work — customers, testing, scaling