How to Migrate from ChatGPT to Claude (Memories and Chats) Easily

Seems like a lot of people are switching from ChatGPT to Claude (or at least starting with Claude and not wanting to lose their histories. This will help.

One of the biggest questions I get from people switching from ChatGPT to Claude is the same every time:

"What do I do with all my memories? How do I bring them with me?"

It's a fair question. You've spent months, sometimes years, teaching ChatGPT who you are, what you do, how you think. Walking away from that feels like leaving a relationship where the other person knows you really well and starting over with someone new.

Good news: you don't have to start over.

Bad news: there's no one-click button that magically moves everything across. At least not yet.

So here's the next best thing. Two methods. One easy. One deeper. Use whichever (or both) fits what you're trying to preserve.

Quick heads up before we start

There is currently no way to import all your individual chats from ChatGPT into Claude as individual chats. That's just the truth of it.

What you can do:

1. Bring over your stored memories (the stuff ChatGPT knows about you)

2. Bring over specific chats you don't want to lose, one at a time, in a usable format

That's what I'll walk you through.

Method 1: The Easy Way (Memory Migration)

This is the fast lane. Maybe 5 minutes of work.

Step 1. Open Claude. Go to Settings, then Capabilities.

Step 2. Find the section called "Import memory from other providers" and click Start Import.

Step 3. Claude gives you a prompt to copy into ChatGPT.

Step 4. Paste that prompt into ChatGPT. ChatGPT will spit out a structured dump of everything it knows about you: identity, career, projects, preferences, the whole picture.

Step 5. Copy that output, come back to Claude, paste it into the field, and hit "Add to memory."

Done. Claude now knows what ChatGPT knew.

One small upgrade I'd add

Before you paste it straight into Claude, drop the ChatGPT output into a Google Doc first and edit it.

Why? Because ChatGPT sometimes gets things wrong. It remembers stuff you didn't actually tell it to remember. It infers things that aren't true. It hangs onto context from one weird conversation three months ago like it's gospel.

Take 10 minutes. Read through it. Delete anything you don't want carrying forward. Then paste the cleaned version into Claude.

Future-you will thank present-you.

Method 2: The Deeper Way (Actual Chat Migration)

This one is for when you have specific chats you can't afford to lose. Maybe a long strategic conversation. Client work. A framework you built collaboratively. Whatever it is, you want the actual chat, not just the summary.

This isn't a one-click move. But it's a great workflow, and as a bonus it'll teach you how to use Claude Cowork from day one.

Step 1: Export your ChatGPT data

Inside ChatGPT, go to Settings, then Data Controls, then Export Data.

Confirm the export. Now wait. This can take 24 to 48 hours depending on how much chat history you have. You'll get an email when it's ready.

When the email arrives, download the zip file.

Step 2: Unzip it and find conversations.json

Open the zip. You'll see a bunch of files. The one that matters is:

conversations.json

If you try to open this in a regular text editor, you're going to have a bad time. It's not built for humans to read. It's one giant blob of JSON.

This is where Claude Cowork earns its keep.

Step 3: Point Claude Cowork at the folder

If you've never used Cowork before, here's the short version: it's the version of Claude that can read and write files on your computer. You point it at a folder, and it can work inside that folder like a teammate would.

Move the unzipped ChatGPT export into a folder you'll remember (mine is just called "temporary" in my Downloads). In Cowork, attach that folder.

Now Cowork can see your conversations.json.

Step 4: Ask Cowork to read the file

I started with a simple prompt:

Cowork went in, read the JSON, and gave me an at-a-glance inventory: 17 conversations, with patterns and themes surfaced.

(Yours might have hundreds or thousands. Doesn't matter. Same workflow.)

It also offered to do any of these next:

• Pull the first prompt of every chat so I could recognize them at a glance

• Apply heuristics (recency, length, repeat themes) and make keep/archive/delete recommendations

• Group everything by theme (business, marketing, personal, etc.)

• Build an interactive HTML triage view I could click through

This is the part most people miss. A folder full of raw JSON is a problem. The same folder, with an AI teammate that can read it, summarize it, sort it, and segregate it, is a much smaller problem.

Step 5: Import the chats you want saved into Claude

Once you've got your keepers exported as individual markdown or JSON files, you bring them into Claude the same way you'd import any file. Drop one into a chat with a prompt like:

Markdown is easier for you to read. JSON is easier for Claude to parse. Either one works. If you have a project in ChatGPT, you can create the matching project in Claude and drop the chats in there.

That's it. You've now got both the memory layer AND the specific chats that mattered to you.

And if you're curious about Cowork specifically, there's a quick-start link here.

Adapt to change. Evolve with intention. Create what's next.

P.S. Have you been looking to get deeper into Claude? Reply back and let me know, I’m planning a training on this.