Can ChatGPT and Claude share the same memory?

Not by themselves: each keeps its own. They can both read from a memory kept outside them, such as New Brain, which each connects to with one sign-in.

Isn't Claude's memory import enough?

It copies what ChatGPT remembers into Claude once. After that each tool learns separately, so the copy goes out of date unless you repeat it, and it doesn't keep where each fact came from.

Do I lose my ChatGPT or Claude history?

No. Both tools keep their own memory and history. You can also bring the export file from either into New Brain, so your past chats are part of your brain from day one.

Why does ChatGPT ask me to confirm each save?

ChatGPT asks before any connected tool writes something on your behalf. It is expected, and New Brain only sees what you choose to send.

One memory for ChatGPT and Claude, not two copies

On Monday you tell ChatGPT the launch has moved to April. On Thursday you ask Claude to draft the launch email, and it writes "March". The two tools keep separate memories, and nothing you tell one reaches the other.

If you run several projects and use both tools, this happens every week. This guide explains why, why copying your memory from one tool to the other only fixes it for a few days, and how to give both tools one memory they read from.

Two memories that drift apart

ChatGPT and Claude each keep their own notes about you: your projects, your preferences, the decisions you mention. ChatGPT calls these saved memories and chat history. Claude has its own memory, kept inside Claude.

The two never talk to each other. A decision you make in a ChatGPT chat stays in ChatGPT. A client name you mention to Claude stays in Claude. And most of what matters never reaches either one, because it happened in a client message, an email or a meeting.

Each tool ends up with its own partial picture of your work, and it gets older every week.

Why copying memory doesn't fix it

Claude has an import tool for this. It gives you a prompt to run in ChatGPT, ChatGPT writes out what it remembers about you, and you paste that into Claude. Plenty of guides walk through it, and it works.

The trouble is that it's a copy, taken once. From the next day, each tool learns new things the other doesn't.

A copy matches on day 1. By day 30 the two tools disagree about the launch date, and each knows something the other doesn't.

You can repeat the copy every week, but then you're the one keeping two memories in sync by hand. And the copy only carries what ChatGPT decided to remember. It doesn't say where a fact came from, so when the two tools disagree you can't tell which one is right.

One memory both tools read

The other way round is to keep the memory outside both tools, in one place, and let each tool read from it and write to it. That's what New Brain does. It's your digital brain: everything you read, said and decided, at work and everywhere else, and every answer shows where it came from.

Here's the example we use. In a ChatGPT chat you agree the launch moves to April, and ChatGPT saves that decision to New Brain. It lands in the Launch project. A week later you ask Claude "when do we launch?" and it answers "April", with "From ChatGPT, 3 Nov" under it. Ask ChatGPT the same question and you get the same answer, from the same source.

Both tools work from one brain. Whatever is saved once can be answered in either, with where it came from.

Because each answer carries its source and date, a disagreement is easy to settle: you can see which message said what, and when.

Set it up in four steps

Bring what you already have. In New Brain, choose Add, then "Or bring your chat history", and pick the export file from ChatGPT or Claude. Your past chats become the starting point, so you don't begin empty.
Connect Claude. In Claude, open the Connectors screen and add New Brain. It takes four steps and one sign-in.
Connect ChatGPT. ChatGPT adds New Brain from its Plugins screen, as a custom server. Where that sits, and which ChatGPT plans have it, has changed during 2026, so the ChatGPT steps on our Connections page are kept current. ChatGPT shows a warning when you add it; that's expected, and New Brain only sees what you send it.
Save as you go. When something is decided, in a chat or anywhere else, save it to New Brain once. ChatGPT asks you to confirm each save.

The full steps for every tool are on the Connections page.

What changes in a normal week

You stop explaining your work to each tool. Whichever one you open, it starts from the same decisions, the same client details and the same way you like things written.

When an AI tool picks up something it isn't sure about, it waits in To check, and you choose Keep, Fix or Remove. Nothing is overwritten: when the launch moves from March to April, the old date stays in the history, marked as replaced, so you can see what changed and when.

If you switch between them, say Claude for writing and ChatGPT for analysis, neither starts from scratch.

What New Brain doesn't do

A few honest limits, so you know what to expect:

It doesn't change ChatGPT's or Claude's own memory. Those keep working as before. New Brain is a separate memory both tools can read.
It only knows what reaches it. A message you never save, and that no connected app brings in, stays out of it.
ChatGPT asks before each save. That's how ChatGPT handles connected tools.

If your week is split across several projects and two or three AI tools, one memory they all read from is less work than keeping their memories in step by hand.

Keep one memory across ChatGPT and Claude, instead of moving it

Key takeaways