fenbs

Portable AI Memory: One Memory Across ChatGPT, Claude and Gemini

Each AI app remembers you separately, and that memory stays with the vendor. There are three ways to make memory portable: copy it between apps, keep it in a tool that sits outside all of them, or keep the work itself on a shared board. Here is what each does and when to use it.

6 min read

Portable AI memory means that what one AI app knows about you, or about your work, is available in the next app you open, so you do not start from zero every time you switch between ChatGPT, Claude and Gemini. Today there are three ways to get it. You can copy memory from one app into another with the import tools Claude and Gemini now offer. You can keep your memory in a separate tool that plugs into all of them, such as MemoryPlugin, AI Context Flow, myNeutron or MemoryLake. Or, for work rather than personal preferences, you can keep tasks, decisions and progress on a shared board that every AI app reads, which is what fenbs does. Most people who use several AI apps for work end up with a personal memory for themselves and a shared record for the work.

Why memory is not portable by default

Each assistant keeps its own memory, inside its own account. OpenAI’s guide to personalizing ChatGPT (opens in a new tab) says memories “let ChatGPT carry useful context from earlier chats into future work,” and that is the catch: it is ChatGPT’s earlier chats and ChatGPT’s future work. Claude and Gemini have their own memories, built from their own conversations. Tell one of them your writing style or your project’s rules, and the others never hear about it. Our posts on ChatGPT project memory and Claude project memory cover what each keeps.

What “portable” should mean

Before choosing a tool, decide what you need it to do. A useful checklist:

  • It works inside every AI app you actually use, not just most of them.
  • You can see everything it holds, edit it and delete it.
  • It stays current: a change made in one app shows up in the others, rather than being copied once.
  • You can tell who wrote each piece: you, a colleague, or which assistant.
  • If several people use it, each person sees only what they should.
  • You can take it with you if you leave the tool.

No single option ticks every box for every kind of memory, which is why it helps to know which kind you need: personal (who you are, how you like things done) or work (what is being done, what was decided, where it stopped).

Option one: copy memory between apps

The simplest move is built in. Anthropic’s help center explains how to import memory into Claude (opens in a new tab): you paste a prompt into your old assistant, it writes out what it knows about you, and you paste the answer under Settings, Memory. Anthropic says imports are experimental and that Claude’s memory focuses on work-related topics, so personal details may not be kept. Google’s help page on importing into Gemini (opens in a new tab) describes the same prompt-and-paste approach, plus uploading exported chat history from ChatGPT or Claude as a .zip file. Google says it is not available in the European Economic Area, Switzerland or the UK, and needs a personal Google account.

This is a one-time copy, not a sync. After the import, each app goes back to remembering separately, and the copy ages. It is the right tool when you are moving from one assistant to another for good; see Claude import memory for what moves and what does not.

Option two: a memory tool that sits outside every app

A portable memory tool keeps your memory on its own service and feeds it into each AI app, usually through a browser extension for the chat websites and an MCP server for apps that support it. MCP is the standard way an AI app plugs into an outside tool; our glossary explains it in plain words. The four most often compared:

These tools are strong at personal context: your preferences, your style, your reference material, and in some cases years of old chats. They stay in sync as you work, which a one-time import does not. Before you install one, read its privacy page: you are trusting it with what you tell every assistant.

Option three: keep the work itself somewhere every app can read

Personal memory does not cover work. When you switch from Claude to ChatGPT halfway through a project, what you need is not “the user prefers short emails” but “these three tasks are open, we decided to drop the March launch, and yesterday’s session stopped at the pricing page.” That kind of memory has to be written by people, read the same way by every assistant, and shared with colleagues.

fenbs is a web app built for that: a simple kanban board where people and every AI app they use keep projects, tasks, decisions and rules, lessons learned and where they left off. ChatGPT, Claude, Claude Code, Codex, Cursor and Copilot connect over MCP and read the same board. On the Google side, fenbs connects to Gemini CLI.

  • Tasks move through To Do, Next Up, In Progress and Completed.
  • People write the Decisions and rules; every assistant reads the rules first, in full, and cannot sign a decision off or pre-approve work.
  • When you say “save progress,” the assistant saves Where we left off for the project, and the next session in any app starts there.
  • History records every change and who made it, a person or a named assistant.
  • Colleagues join a team board with a role, so each sees what that role allows.

fenbs does not remember your personal preferences or search your old chats. It is the work half of portable memory, not the personal half. For teams, AI memory for teams goes further.

The two halves also fail differently. A personal memory that goes stale costs you a slightly worse answer. A work record that goes stale costs a colleague a day, because they redo a finished task or undo a decision nobody told their assistant about.

Which to use

  1. Leaving one assistant for another for good: use the built-in import, once.
  2. Using several assistants for yourself and tired of re-explaining who you are: pick one personal memory tool.
  3. Doing work that other people, or other assistants, have to pick up: keep it on a shared board.
  4. Building an AI app that needs memory for your own users: none of these. Look at a developer memory engine, starting with what Mem0 is.

Make your work portable

fenbs is free to start. Connect it in Claude or ChatGPT as a custom connector, and every assistant you add reads the same board. How it works shows the pieces, and AI agent hand off explains Where we left off.

Questions people ask.

Can I have one memory across ChatGPT, Claude and Gemini?

Not natively. Each app keeps its own memory. You can copy memory into Claude or Gemini with their import tools, which is a one-time copy, or use a separate memory tool that plugs into all three through a browser extension or MCP and stays in sync.

Is importing memory into Claude or Gemini the same as syncing it?

No. An import copies what your old assistant wrote out about you at that moment. Afterward, each app remembers separately, so anything new you tell one does not reach the other.

What is the difference between portable personal memory and a shared board?

Personal memory carries facts about you, such as preferences, style and past chats, between AI apps. A shared board like fenbs carries the work: tasks, decisions and rules, and where you left off, which people write and every connected assistant reads and updates.

Is portable AI memory safe?

It depends on the tool. A portable memory lives on another company’s service, so read its privacy page, check that you can see, edit and delete what it holds, and check what it sends from your browser before you install it.

Related articles

Start with one thing.

Create a kanban boardHow it works

There is nothing to set up first. Write one line and you’ve started.