A few years ago, I worried that AI might solve coding.

Now I think that question was too small.

If coding and mathematics become abundant, the scarce resource moves elsewhere:

What is worth building?

That is business philosophy: choosing the problems, values, and tradeoffs behind a product.

AI can make execution cheap. Judgment still matters.

These are the three principles I use.

1. AI-Native Is a Cliché. Change the Ontology.

My test is simple:

If the AI really works, what becomes unnecessary?

“AI-native” tells me almost nothing. Waymo depends on AI. So do thousands of wrappers.

I care about businesses where AI changes what the product fundamentally is.

A button disappears. A workflow disappears. A human operation disappears. Sometimes even a physical component disappears.

A Cybercab without AI is not a worse car. Without a steering wheel or pedals, it becomes a useless box.

That is the distinction.

AI should not merely automate the old ontology. It should make part of the old ontology obsolete.

2. Efficiency Is Cheap. Build Human Flourishing.

Completing tasks faster is not enough.

I care about businesses that expand human agency: giving people the ability to create, find an audience, earn, and build something that feels like their own.

YouTube and Steam did more than distribute content. They created ecosystems where ordinary people could turn creativity into recognition, independence, and sometimes a livelihood.

AI can make creation dramatically easier. But more output alone means little if people gain no ownership, audience, reputation, or income from it.

My test is:

Does this product make people more replaceable—or give them more power over their own lives?

Technology should not merely increase productivity. It should expand the possibilities of human flourishing.

3. Models Commoditize. Moats Must Compound.

Better models help you—and everyone else.

A real moat comes from value that compounds through use: relationships, reputation, distribution, trust, data, or network effects that competitors cannot simply download.

For a creator platform, the moat is not the content library. It is the web of creators, audiences, discovery, reputation, and economic relationships built around it.

My test is:

If everyone gets the same models tomorrow, why do people still choose us?

AI should improve the product. Participation should make it harder to replace.

The strongest businesses create their own gravity: creators attract audiences, audiences reward creators, and every interaction makes the ecosystem more valuable.

Together, the three principles form a sequence:

Change the ontology → Expand human agency → Build compounding gravity

Why can this exist now?
Does it give people more power?
Why does it become harder to replace?

Principles Are Cheap. Backtest Them.

It is easy to explain success after it happens.

Google won, so every Google trait suddenly looks profound. Failed companies disappear from memory. Hindsight turns accidents into theories.

That is why I am building Dotcom Déjà Vu.

The question is deliberately uncomfortable:

Could we have recognized the next Google before everyone knew it was Google?

The rule is simple:

No hindsight.

Compare Google with AltaVista using only what was knowable then. Compare Facebook with Friendster at the moment the choice was still uncertain.

The goal is not to manufacture a prophecy machine. It is to expose the assumptions behind judgment—and see which ones survive contact with history.

A philosophy that only explains winners afterward is not a philosophy. It is storytelling.

Intelligence Is Abundant. Judgment Is Not.

AI can code, analyze, design, search, and increasingly reason.

So what remains valuable?

Judgment.

Choosing which problems deserve solving. Deciding which assumptions to destroy. Knowing what should be optimized—and what should never be optimized at all.

AI can participate in those decisions. It cannot make technical capability itself tell us what is worth building.

That is where philosophy survives.

Math may be solved. Coding may be solved. Philosophy is not.