Transfer ChatGPT memory without losing three-plus years of conversation history. Here’s exactly how.
How to Transfer ChatGPT Data Step by Step
You’ve probably built up months (or years) of useful context inside ChatGPT. Custom instructions. Writing style preferences. Business context. It’s sitting in a 247MB JSON file you’ve never looked at.
If you’re moving to Claude, Gemini, or Google NotebookLM, you don’t have to start from scratch. You just need to know how to get that data out in a usable format.
If you’re serious about AI workflows, my free guide covers the exact prompts, systems, and templates I use every day.
Free AI Marketing Essentials Guide (includes my Claude prompts, automation templates, systems, and more).
Step 1: Export Your ChatGPT Data
Go to ChatGPT and click your profile icon in the bottom left. Select Settings, then Data Controls. You’ll see an Export Data button. Click it, confirm, and OpenAI will send a download link to your email.
Open that email and click Download Data Export. A zip file lands in your Downloads folder. On Mac, double-click it to unzip automatically. Once it’s done, rename the unzipped folder to exactly chatgpt data (no caps, no hyphen). That name matters for the next step.
Inside that folder, you’ll see a file called conversations.json. That’s your entire chat history. The problem? It’s massive. At 247MB, no AI platform can process it as a single file.
Step 2: Split the JSON Into Usable MD Files
This is where most guides stop. They tell you to export but don’t explain what to do with a 247MB file that won’t upload anywhere.
The fix is a one-shot terminal command that splits conversations.json into separate Markdown files. Credit to Corey McClain for the original command. You can find it linked in Ryan’s video description.
Open Terminal on Mac (Windows users will need to find an equivalent). Paste the command and hit return. When it finishes, you’ll see done in the terminal output, and a new folder called notebook lm ready will appear inside your Downloads folder.
That folder contains 14 separate MD files covering your full conversation history. Drag it to your desktop for easy access in the next step.
Step 3: Upload to NotebookLM
Here’s why NotebookLM is the right starting point instead of going straight to Gemini or Claude.
NotebookLM extracts data from your sources verbatim. It doesn’t mix in outside training data or internet information. When you ask it something, it pulls answers directly from your uploaded files. That matters when you want to analyze your own conversation history without AI hallucinating additions.
Open NotebookLM and create a new notebook. Name it something like ChatGPT Data. Click Add Sources, then Upload Sources, and select all 14 MD files from your desktop folder. Upload time varies based on file size, but it’s manageable.
Once they’re uploaded, you can start chatting with your own history. A simple prompt like “Act as a chatbot that perfectly mimics the voice, tone, and style of the person from the sources” gets you going immediately. NotebookLM will cite specific conversations as it responds.
You can also use NotebookLM’s other output formats: audio overviews, infographics, slide decks, flashcards. Your chat history becomes a fully searchable knowledge base.
Step 4: Connect NotebookLM to Gemini or Claude
NotebookLM’s built-in chat runs on Gemini Flash, not the more advanced Gemini Thinking or Pro models. To use your full chat history with a stronger model, you need to link it.
Inside Google Gemini, click Add Files, then select NotebookLM. Find the notebook you just created and click Add. Switch your model to Gemini Thinking or Pro. Now you’re running a top-tier model against your full ChatGPT history.
One practical limit: Gemini accepts up to 10 file attachments per chat window. NotebookLM sidesteps that cap, which is the real reason to route through it first.
For Claude, the process is more direct. Claude accepts all 14 MD files at once without hitting Gemini’s 10-file limit. Create a Claude Project, upload your files, and you’re done.
What can you actually do with this? One high-value use case: paste a Brand DNA construction prompt into Gemini with your notebook attached. Gemini will generate a personal brand style guide based on three-plus years of your writing. Tone, structure, word choice, reading level. All pulled from your actual work, not guesswork.
That output is useful for training AI agents in Claude, building automation modules in Make or Zapier, or setting up a Google Gem that writes in your voice from day one.
Step 5: Export Claude Data Too
Worth noting: you can run this same export from Claude. Click your icon in the bottom left, go to Settings, then Privacy, and select Export Data. Claude will email you a zip file with your full conversation history.
The same terminal command that splits ChatGPT’s JSON will work on Claude’s export. If you’re a heavy Claude user and want to move workflows to Gemini or a new tool, this is how you do it without starting over.
Final Thoughts from Ryan
The point isn’t to abandon ChatGPT out of frustration. The point is that your conversation history has real value, and you shouldn’t leave it locked in a platform you’re using less.
Three-plus years of context, writing patterns, business decisions, and problem-solving is sitting in that JSON file. Once you move it into NotebookLM, Gemini, or Claude, it becomes an active asset instead of a forgotten archive.
The process takes about 20 minutes. The terminal command is the only mildly technical part, and it’s a copy-paste job.
Transfer ChatGPT Memory FAQs
Can you transfer ChatGPT memory automatically?
No automatic tool exists for this yet. The process requires manually exporting your data from ChatGPT’s settings, splitting the conversations.json file using a terminal command, and uploading the resulting MD files to NotebookLM, Claude, or Gemini. It takes around 20 minutes and is mostly copy-paste work.
How do I transfer ChatGPT data to Claude?
Export your ChatGPT data from Settings > Data Controls > Export Data. Unzip the file, rename the folder to “chatgpt data,” and run the one-shot terminal command to split conversations.json into MD files. Then create a Claude Project and upload all 14 MD files directly. Claude accepts the full batch without a file cap.
How do I transfer ChatGPT memory to Gemini?
Follow the same export and split process, then upload the MD files into a NotebookLM notebook. Inside Google Gemini, click Add Files > NotebookLM, select your notebook, and switch to Gemini Thinking or Pro. This routes around Gemini’s 10-file-per-chat limit and gives you access to stronger models.
How do I transfer ChatGPT memory to NotebookLM?
After exporting and splitting your ChatGPT data into MD files, create a new notebook in NotebookLM and upload all files as sources. NotebookLM will index everything verbatim, letting you query your full conversation history without mixing in outside training data.
What is the conversations.json file in ChatGPT data export?
It’s a single file containing your entire ChatGPT conversation history. It’s typically too large (200MB or more) to upload directly to any AI platform. The terminal command splits it into smaller MD files that NotebookLM, Gemini, and Claude can each process.
