How to switch From ChatGPT to Claude

I get this message from subscribers at least twice a week. Some versions of: "I tried switching to Claude from ChatGPT. Gave up after two weeks."

Every time I ask the same question: did you actually set it up?

The answer is always no.

And that's the whole problem.

Switching tools without changing the setup is like moving to a new kitchen and trying to cook the same meal the same way, except every appliance is somewhere different and the knives are sharper than you expect. It works eventually. But the first week is frustrating enough that most people order takeout and call it a loss.

This guide is the setup. Five steps. Everything you need to bring your history over, rebuild your memory, and use Claude the way it was actually designed to be used.

When you've used ChatGPT for two years, you've built up something real.

Hundreds of conversations where you explained how you think. What you're building. What format do you like? What you rejected. Hundreds of times you gave feedback and iterated toward something that finally worked.

That context feels like lost work when you switch. It's not. You can bring all of it.

The actual problem is something different. Most people who switch bring their ChatGPT workflow with them and expect the same results:

→ New chat every session → Context copy-pasted at the top every time → Manual iteration on every output → One tool, one surface, everything through the same window

That workflow was designed around ChatGPT's limitations. Claude was built around Skills, Projects, Connectors, and memory that loads automatically. Using Claude the ChatGPT way is like using a dishwasher to hand-wash one plate. It works. But you're missing the point entirely.

The people who switch and stay switched rebuild their workflow inside Claude. This is how.

ChatGPT lets you export everything. Every conversation, every instruction, every piece of context you've built up. Most people don't know this exists.

Here's how:

Inside the folder you'll find several files. The one that matters is chat.html.

That file is your entire conversation history. Every prompt you've ever sent, every response you received, every piece of context you've accumulated across however long you've been using it.

Go to Settings → Personalization → Custom Instructions. You'll see a field:

→ "Additional behaviour, style and tone preferences"

Copy it. Paste them somewhere you can access in a few minutes.

That text becomes your Claude User Preferences in Step 3. Don't skip it.

Open Claude. Create a new Project. Name it after your actual work area. This project is going to be your permanent home base.

Inside the project, find Instruction and upload chat.html.

That's the whole step.

Here's what just happened. Every conversation you've ever had in ChatGPT is now a permanent context inside that project. Not a file you re-upload. Not a block of text you paste into every chat. Knowledge that loads automatically into every conversation you open inside this project.

It's like handing a new assistant every meeting note, every email thread, every decision log from the last two years, before they sit down for their first session with you. They don't need the backstory. They already have it.

If the answer is accurate, the upload worked. If it's off, add a short project instruction giving Claude more context about your main focus. Two or three sentences is enough.

Become a paying subscriber to get access to this post and all other premium content.