Every Industry Will Eventually Split Into Three Types of Companies

A while ago, I had dinner with a friend who runs a home renovation business.

He complained to me: customer acquisition keeps getting harder. He's advertised on Xiaohongshu (RED, China's leading lifestyle social platform), shot videos on Douyin (TikTok's Chinese sister app), kept a sales team on payroll — busy from one year to the next, with margins as thin as paper.

I asked him, are you using AI?

He froze for a second. Said they'd tried it — had the young woman on the team use a chatbot to write a few pieces of copy. Didn't help, so they let it go.

I didn't say anything.

But my stomach dropped a little. Because I suddenly realized something —

Every company, in every industry, may eventually split into three types.

The First Type: "AI-Native"

What does AI-native mean?

It means that from the very first day this company was registered, its product, its processes, its people — all of them grew up around AI.

Without AI, this company wouldn't exist at all.

Think about how frightening that is. Its cost structure isn't even on the same ledger as yours. What ten people there produce might match fifty of yours. A new feature there might ship in two days, while you're still scheduling it.

It cuts to the bone. It's not outworking you — it's playing a different game entirely.

The Second Type: "AI-Transformed"

This kind of company was doing perfectly fine. It had customers, a team, accumulated experience.

But it made a decision: to truly internalize AI.

Not "internalize" in the write-a-couple-pieces-of-copy sense — but to jam AI into its analysis, its ad placement, its customer service, into every single corner where it could be jammed.

I saw a take recently that really stuck with me: it used to take weeks to put together a competitive analysis — the research, the organizing, the back-and-forth review. And now? With AI, you can spin up a decent first draft in maybe a minute.

One minute.

Let that number sink in. That's not "a little faster" — it crushes you by orders of magnitude.

What's the hallmark of this kind of company? It already had an established foundation, and AI scaled it up. It isn't starting from zero. It's bringing its customers, its data, its industry understanding — and then adding AI as leverage on top.

Once it gets rolling, it's even fiercer than the AI-native ones. Because it has reserves to draw on.

The Third Type — You Can Probably Guess

Elimination.

That home renovation friend of mine — is he in danger?

I don't think it's come to that yet. But he's standing at the edge of a cliff.

Why?

Because among his competitors, the moment a single one starts using AI to run customer acquisition, to spin up design drafts, to follow up with clients — once their costs get pushed down a notch — those "thin as paper" margins of his will genuinely be gone.

Elimination doesn't happen in an instant. It's the boiling-frog scenario. This year you feel okay; next year you feel the strain; the year after, you suddenly realize the clients, for some reason, just aren't coming anymore.

Clients won't tell you where they went. They just stop coming.

The Real Question Isn't "Whether to Use AI"

A lot of people are still tangled up over this one.

Should we use AI? Which tool? Which model is best?

These are the wrong questions.

The one worth asking is: in your team, is there anyone — systematically, persistently — turning AI into the tool that actually does the work?

I once heard a very real worry: a marketing team of a dozen-plus people, where two or three are really into tinkering with AI and have gotten insanely good at it. The rest basically don't touch it.

Give it time — what happens?

Those two or three, their output shoots up. Everyone else, standing still. The internal "AI gap" on the team opens up, little by little, just like that.

Once the gap widens to a certain point, output can't keep up. When it can't keep up, heads roll — or the market does it for you.

I'm not trying to scare anyone. This is just the math.

And There's Another Ledger That Cuts Even Deeper

AI isn't free.

The deeper you go, the more you spend. Model calls cost money. Building workflows costs money. Keeping someone on staff who actually gets it costs even more.

I've seen plenty of companies get incredibly excited at the start — wanting to try everything, buying every tool. After a lap of experiments, they've burned through a pile of budget, and the use cases that actually stuck amount to almost none.

That's the awkward spot.

You spent money testing a pile of things that went nowhere — and when you look up, the budget's gone, but your competitors weren't sitting still.

So what's the smarter move? Hold back. Don't try everything. Pick the single most painful link in your chain and get that one running end-to-end first. Once it works, then expand.

Less is more. On the matter of AI, that's especially true.

One Thing Matters More Than Saving Money

Have you ever thought about this — the stuff you've got sitting inside some AI tool right now, your prompts, your workflows, the templates you've tuned — what happens the day that tool stops cutting it, or you switch?

Models turn over fast. What's great today might not be tomorrow.

The one you depend on might shut down one day, or get pricier, or stop being good.

So what you really need to protect isn't any one tool. It's your judgment, and the methodology you've built up along the way.

Tools will be swapped out. But knowing "how to use AI to solve this kind of problem" — that doesn't get swapped out.

That's your real asset.

One Last Feeling, From Me

I've come to feel more and more that AI isn't a technology problem.

It's a question of "are you willing to take yourself apart, and reassemble yourself, piece by piece."

The ones who are willing — even if they've only put themselves halfway back together — are already getting a taste of the payoff. The ones who aren't are still standing in place, waiting for the "perfect moment," waiting for the moment when they've "fully understood it before they act."

But the market doesn't wait.

While you're taking your time looking, someone else is already running. By the time you've finally figured it out, your clients, your people, your margins — they may all be somewhere other than where they started.

Every industry will eventually split into three types of companies. The only question is which one you intend to be.

I'll leave that question right here.

The answer is one you'll have to give yourself. Just don't give it too late.