Everyone is about to rebuild their AI setups

Last week I saw people panic.

Not loudly. But you could feel it.

Anthropic raised Claude’s API pricing overnight.

Cheap third-party usage?

Gone.

Everything is moving to pay-as-you-go.

And suddenly…

A lot of AI setups stopped making sense.

This always happens.

A platform changes pricing. An API gets restricted. A tool disappears.

And the people who built everything on top of it…

Start over.

I’ve been there.

Building workflows around one model.

Optimizing everything for it. Getting it “perfect”.

Until one small change…

And the whole thing breaks.

That’s when it clicked for me:

AI isn’t one tool.

It’s a stack.

Now I build differently.

Each model has a role.

Claude Code → coding, because it understands full repos and ships clean code.

GPT 5.4 Pro → agents, because it handles complex reasoning chains best.

Minimax M2 → background jobs, because it’s fast and cheap for bulk tasks.

Nothing overlaps. Nothing depends on just one piece.

So when something changes…

I don’t rebuild.

I swap. 

Most people are still thinking in tools.

“What’s the best model?” “What should I use?”

Wrong question.

The real question is:

“How do I design this so it doesn’t break?”

Because this won’t be the last change.

It’s just the next one.

If your current setup would fall apart because of one pricing update…

You don’t have a system yet.

You have a dependency.

Need help or have questions? Just reply to this email.

P.S. Interested in getting coached to your goal? Check out my coaching packages here:

Click here