I’ve lost count of how many people asked me some version of this lately: “I want to try the other one, but I don’t want to lose everything my AI already knows about me.” That fear is real. Your chat history, your memories, your custom setup, it all feels locked inside one app.
So I was genuinely happy to stumble on a post from a LinkedIn creator who tackled exactly that. The original poster’s take is refreshingly blunt: Claude or ChatGPT, they’re about the same. What actually matters is how painlessly you can move between them. And it turns out the switch takes a handful of steps in either direction.
I’ve pulled the process apart below, with the reasoning behind each step, so you can follow along without guessing.
Why this matters right now
The author paints two scenarios most of us will recognize. Someone in your circle just said, “I love the new GPT-6 Astra, but how do I switch from Claude to ChatGPT?” Or a friend has been using ChatGPT for personal questions and now wants Claude only for professional work. Both are legit. And both used to mean starting over with an assistant that knew nothing about you.
The clever part of this approach is that it treats your memories and history as portable data, not something welded to one vendor. Once you see it that way, the whole migration stops feeling scary.
Route 1: Moving from Claude to ChatGPT
This direction has one extra hurdle, because Claude’s export gives you raw files, not a tidy memory summary. The author’s workaround is smart.
- Export your data. In Claude, go to Settings, then Privacy, then Export data, and pick the 90-day window. Why: this pulls your recent conversations and memories into a package you actually own.
- Grab the zip. Claude emails it to you. Download it and unzip it. Why: the export arrives as a bundle, and you need the files loose to upload them.
- Turn it into one Markdown file. Open a fresh Claude chat, upload the unzipped files, and run a conversion prompt. The creator wrote one that turns your entire history plus memories into a single file called Claude-to-ChatGPT.md. Why: ChatGPT can’t make sense of a pile of exports, but it reads one clean, structured document just fine.
- Feed it to ChatGPT. Upload that Markdown file to your ChatGPT Library and ask it to save the contents to memory. Why: this is the step that actually teaches ChatGPT who you are.
If you want to try step 3 with your own wording, a plain version works: “Read every file I uploaded. Summarize my preferences, projects, recurring topics, and anything I asked you to remember into a single Markdown document I can hand to another AI assistant.” Not as polished as the expert’s version, but it gets you most of the way.
Route 2: Moving from ChatGPT to Claude
Going this way is smoother, because Claude has a built-in memory import flow. The post’s author lays it out in five moves.
- Open the memory import screen. In Claude, head to Settings, then Capabilities, and look for the memory import option. Why: Claude generates a tailored export prompt for you right there, so you don’t have to write one.
- Copy the export prompt. Claude hands you a ready-made prompt. Copy it exactly as it is. Why: it’s built to pull memories out in a format Claude already understands.
- Paste it into ChatGPT. ChatGPT responds with all your memories packed into one code block. Why: the code block keeps the formatting intact for the return trip.
- Bring it back to Claude. Paste the code block into Claude and choose Add to memory. Why: now Claude carries everything ChatGPT knew about you.
- Move your skills too. Find your Codex Skills folder, download each skill as a zip, then in Claude go to Customize, then Skills, then Add. Why: memories cover context, but skills cover workflows, and you don’t want to rebuild those by hand.
Why it matters: the real switching cost with AI assistants was never the subscription price. It was the months of context you’d built up. Once memories and skills are portable, you can pick a model based on how well it does the job today, not on how much you’ve already invested.
Tips I’d add from my own experience
- Clean before you migrate. Skim the export first and drop stale projects or outdated preferences. A migration is the perfect excuse to declutter.
- Keep the Markdown file. That single document is a portable backup of your AI context. Stash a copy somewhere safe.
- Test with a question only your old assistant could answer. Ask the new one about a project you never mentioned there. If it answers correctly, the memory took.
- Run both if you’re undecided. Loading the same context into both platforms is the fairest side-by-side test I know of.
Where this gets really useful
The split the author describes, personal questions on one platform and professional work on the other, is more practical than it sounds. Your work assistant doesn’t need to know your meal plans, and your personal one doesn’t need your client list. Migrating just the relevant slice keeps each assistant focused and keeps sensitive stuff where it belongs.
It’s also a quiet reminder that these tools are closer to interchangeable than the marketing suggests. When a new model drops and everyone rushes to try it, the people who move fastest are the ones who already know their context travels with them. I think that’s the mindset shift worth taking from this post!
The original poster shares the full walkthrough with all the details, so head over to LinkedIn and check out the complete post before you start your own switch.