How to Transfer ChatGPT Memory to Claude
A lot of people are moving from ChatGPT to Claude right now, and the biggest barrier is losing all the context ChatGPT has built up about you over months or years of use. Your custom instructions. Your stored memories. The way it's learned how you write, what you work on, the tools you use. Starting
A lot of people are moving from ChatGPT to Claude right now, and the biggest barrier is losing all the context ChatGPT has built up about you over months or years of use.
Your custom instructions. Your stored memories. The way it’s learned how you write, what you work on, the tools you use. Starting from scratch with a new AI assistant feels like training a new colleague who knows nothing about you.
So here’s a complete guide on how to transfer your ChatGPT memory to Claude, step by step. I’ve put together a two-part prompt that extracts significantly more context than the default method, and I’ll walk you through the entire process from export to import.
Why people are switching to Claude
Two big reasons.
First, Anthropic’s standoff with the US Department of War. The Pentagon gave them an ultimatum: remove your restrictions on Claude being used for mass surveillance of Americans and fully autonomous weapons, or lose your government contract. Anthropic said no. They got blacklisted, designated a supply chain risk, and replaced with OpenAI, all within days.
You can have whatever opinion you want on that, but a company holding a line that costs them hundreds of millions of dollars when they could have quietly caved has earned a lot of respect.
Second, Claude has just been getting really good. The updates have been coming thick and fast, and for a lot of day-to-day work, writing, analysis, coding, reasoning, it’s become my go-to. If you want to make Claude’s output even better once you’ve switched, I’ve written a guide on how to stop Claude writing like an AI that’s worth reading alongside this one.
What you’ll need before you start
This process takes about 5 minutes. Here’s what you need:
- An active ChatGPT account (free or paid)
- A Claude account (free accounts now support memory as of March 2026, though paid plans give you more capacity)
- A text editor or notes app for cleaning up the export (Google Docs, Notion, Apple Notes, whatever you use)
Step 1: Export your ChatGPT memory and custom instructions
Open a new chat with ChatGPT and paste in this first prompt. It pulls out your stored memories and custom instructions in one clean block.
Prompt 1: Extract stored data
I'm moving to a different AI assistant and I need to export everything you know about me. Please provide ALL of the following in a single, clearly structured response:
1. STORED MEMORIES
List every single memory you have stored about me. Write them out verbatim, exactly as they appear in your memory. Don't summarise or paraphrase — I want the raw entries.
2. CUSTOM INSTRUCTIONS
Reproduce my full custom instructions exactly as written — both:
- "What would you like ChatGPT to know about you?"
- "How would you like ChatGPT to respond?"
3. RECURRING PATTERNS
Based on our conversation history, list:
- Topics I ask about most frequently
- Types of tasks I use you for most often
- Any tools, platforms, or technologies I regularly mention
- My apparent role, industry, or professional context
- Any preferences you've noticed about how I like responses formatted
4. COMMUNICATION STYLE
Describe my writing/communication style as you've observed it:
- Tone (formal, casual, technical, etc.)
- Typical message length
- Any vocabulary or phrases I use often
- Languages I communicate in
Please be exhaustive. Include everything, even if it seems minor. I want a complete picture of what you've learned about me.
Prompt 2: Extract conversation patterns
Once ChatGPT has responded, follow up with this second prompt in the same chat:
Now I need you to go deeper. Look across all our conversations and identify:
1. PROJECTS AND GOALS
- Any ongoing projects or goals I've mentioned
- Business or career objectives I've discussed
- Personal goals or interests that have come up
2. DECISIONS AND PREFERENCES
- Tools or approaches I've said I prefer
- Things I've said I dislike or want to avoid
- Any strong opinions I've expressed about how things should be done
3. CONTEXT THAT MATTERS
- My team structure or colleagues I've mentioned
- Clients, companies, or organisations I work with
- Recurring deadlines, schedules, or rhythms in my work
4. CORRECTIONS AND REFINEMENTS
- Times I've corrected your output or asked you to adjust
- Specific feedback I've given about what works and what doesn't
- Any "rules" I've established for how you should handle certain tasks
Again, be as thorough as possible. I'd rather have too much than too little.
Between the two prompts, you’ll get a comprehensive picture of everything ChatGPT knows about you. This two-part approach extracts significantly more than Anthropic’s single export prompt, because it forces ChatGPT to look beyond its stored memories and into the patterns from your actual conversation history.
Step 2: Clean up and edit your export
This is the step most people skip, and it’s the most important one.
Paste everything ChatGPT gave you into a document and go through it. Strip out anything that’s:
- Outdated: old tasks you’ve finished, projects that no longer exist, tools you’ve stopped using
- Irrelevant: one-off conversations you never came back to, random trivia
- Too personal: anything you don’t need an AI to know about you
- Inaccurate: ChatGPT sometimes infers things incorrectly from context, so check everything
Think of this cleaned-up version as your portable AI profile. You could take it to Claude, Gemini, or anywhere else. A well-edited profile is more useful than a bloated one, so don’t be afraid to cut aggressively.
If ChatGPT’s response is thin, it probably means you don’t have memory switched on, or you haven’t been using it long enough to build up much context. In that case, the manual route works too: just write up your own profile covering the points above. Honestly, a hand-written profile is often more useful than an auto-generated one anyway.
Step 3: Import your memory into Claude
There are two ways to get your profile into Claude.
Option A: Use Claude’s built-in import tool (recommended)
Go to claude.com/import-memory, or in Claude’s settings navigate to Settings > Capabilities > Memory Import.
Paste your cleaned-up profile into the text box and click “Add to memory.” Claude will process it into its memory system.
It can take up to 24 hours for everything to fully integrate, so don’t worry if it’s not all there immediately.
Option B: Paste into a new chat
Start a new conversation with Claude and paste your profile with this instruction at the top:
This is my profile exported from another AI assistant. Please add all of this information to your memory so you can reference it in future conversations:
Claude will confirm what it’s stored. This method gives you more control over what goes in, because you can have a conversation about it.
Check it worked
To verify the import, start a fresh chat and ask Claude: “What do you know about me?” You should see your key details reflected back.
You can also review and edit what Claude has stored by going to Settings > Capabilities > View and edit your memory.
What about Custom GPTs?
If you use Custom GPTs, their instructions won’t be included in the memory export. You’d need to recreate those as Claude Skills or Projects, depending on how you use your custom GPTs.
Copy your GPT’s instructions into a Project’s system prompt, or if you want to recreate it as a skill, paste in all the information from your GPT into a new Claude chat and ask it to create a skill based on the information.
What about exporting your full chat history?
If you want the nuclear option, you can export your entire ChatGPT conversation history via Settings > Data Controls > Export Data. OpenAI will email you a zip file.
This is overkill for most people, but if you want to do a detailed analysis of years of conversations, it’s there. You can upload the relevant files to a Claude Project for analysis, though be aware of file size limits.
Making the most of Claude once you’ve switched
Once your memory is imported, Claude will start personalising its responses based on what it knows about you. But there are a few things worth doing to get the best experience:
Set up your custom instructions. Claude has its own equivalent in Settings > Profile. Use this to tell Claude how you want it to respond: tone, format, length, anything that should apply to every conversation. If you had good ChatGPT custom instructions, here’s how to think about custom instructions and what makes them effective.
Explore Projects. If you used ChatGPT for specific recurring tasks (client work, content creation, coding), set up Claude Projects with tailored system prompts for each one.
Try the connectors. Claude integrates with tools like Google Drive, Notion, Slack, and others through its connectors feature. Worth exploring if you want your AI assistant connected to where your work actually lives.
Frequently asked questions
Can I transfer ChatGPT conversations to Claude? Not directly. Claude’s import tool transfers your memory and preferences, not your full chat history. If you need to reference old conversations, export your ChatGPT data and upload specific files to a Claude Project.
Does Claude’s memory import work with the free plan? Yes. As of March 2026, Claude’s memory feature is available on all plans, including free. Paid plans (Pro, Max, Team, Enterprise) offer more memory capacity and features like Projects.
Can I import memory from Gemini or other AI assistants? Yes. The same process works with any AI that stores context about you. Run the export prompts in Gemini, Copilot, or whatever you’re using, clean up the output, and import it into Claude.
How do I export my Claude memory if I want to switch again? Go to Settings > Capabilities > View and edit your memory, or ask Claude in a conversation: “Write out your memories of me verbatim, exactly as they appear in your memory.” Copy the output and save it.
What if Claude didn’t retain something I imported? Claude’s memory prioritises work-relevant context. If something important didn’t stick, you can add it manually in Settings > Capabilities > View and edit your memory.
Is the memory import feature permanent? Anthropic currently labels it as experimental, but it works reliably for most users. Manual editing is always available as a backup.
That’s how to transfer your ChatGPT memory to Claude. The whole process takes about 5 minutes, and the result is an AI assistant that knows how you work from the first conversation.