I Gave My Businesses a Board of Directors. They're All AI.

This blog is an experiment with a simple question: can AI advise and build a real business — not in a demo, but in the real world, with a human owner directing it and real money on the line?

I'm not a spectator in it. I own the businesses, I make the final calls, and my name is on the results — profitable or not. These are early-stage projects, pre-revenue as I write this, with less than $500 invested so far in domains, tools, and setup costs. But I don't make those calls alone. I have a board.

How the board was created

The board started as a practical problem. I was using several AI tools at once — the subscriptions I already pay for — and getting a different answer from each one. Instead of picking whichever answer sounded best, I wired them together into a panel and made them answer the same question independently, on the record.

That's the board: multiple frontier AIs — ChatGPT, Claude, and Gemini — seated as advisors, with an orchestrator agent (Chaos) who briefs them, runs the sessions, and reconciles what they say into one recommendation. Where they agree, that's a signal to dig deeper — not proof they're right, since models can share blind spots. Where they disagree, the disagreement gets written down, not smoothed over. Then I decide.

It's not a toy. The board has already influenced decisions about where I'll spend real time and money — three sessions in the past two weeks:

• Asked which side hustle deserved my time, all three advisors independently picked the same one — a local home-services directory and lead-generation business I'm building in Des Moines.
• Asked to rank five directory niches, they unanimously put radon mitigation first — a pick I would not have made on my own. I then checked the underlying market assumptions before accepting it.
• Asked about my digital-product pricing, they told me it was indefensible against the competition. Their criticism held up when I compared the plan against current listings, so I changed the pricing strategy.

Three different AIs, three different companies. They can still share blind spots — but comparing their reasoning makes weak assumptions easier to spot. That last part is the point.

What the board is for

The board exists to make the decisions behind this blog's one subject: creating and running businesses with AI.

It pressure-tests business ideas before money goes in. It helps me score opportunities and kill weak ones early, and it reviews every post on this blog before it publishes. And it holds the projects — and me — to account: this blog runs on receipts. Revenue, costs, and hours get published monthly, including the failures. If the board calls a shot and it misses, the miss goes on the record too.

How the board will evolve

The board you meet today is version one, and it's supposed to change:

• Seats will rotate. Models improve and new ones ship. Any advisor that stops earning its seat gets replaced, and the change gets documented here.
• Its job will grow. Today the board advises and reviews. Over time, more of the execution moves to AI agents working under supervision — building, monitoring, and reporting — while the board's role shifts toward oversight.
• Its calls get graded. Every significant recommendation gets a dated follow-up within 90 days — progress or an explanation, not silence. A board that never has its record checked is just decoration, so the scorecard is part of the format.
• The agenda will widen. The businesses come first, but reader questions and reader-submitted ideas are planned for future sessions.

The human part

One rule does not change as the board evolves: a human reviews everything. Every post, every recommendation, every dollar. I read what the board produces, I make suggestions, I send things back for changes, and I can overrule any of it — because the businesses, the money, and the consequences are mine, not theirs.

The board is artificially run. The responsibility isn't.

One more thing: I know what this looks like. "Make money with AI" blogs are everywhere, most of them selling courses on businesses their authors never built. This one publishes the receipts — revenue or lack of it, hours invested, what worked and what didn't — on a schedule, before asking for a dollar. That's the deal.

Next post: the board's first big session — I asked three AIs which side hustle could actually replace a salary, and they answered with one voice. What they picked, what they cut, and what I'm doing about it.

— Anthony

This post was drafted by AI, reviewed by the board, and approved by me before publication. That's the format for everything here.