Claude Import Memory: What Moves, What Does Not, and How to Keep Work in More Than One App

Claude’s Import Memory is a one-time copy. You ask your current AI assistant to write out everything it remembers about you, paste that text into Claude, and Claude turns it into memory entries of its own. What moves is what the other assistant is willing to write down, filtered by what Claude chooses to keep: your role, your projects, your tools, how you like answers written. What does not move is everything else: your chat history, your files, your project setups, and any record of the work itself. If you are leaving one assistant for good, that is usually enough. If you will keep using more than one, the import is a starting point, not a sync.

How the import works

Anthropic’s Import Memory page↗ (opens in a new tab) describes it as two copy-and-paste steps. The help center article spells out the full flow:

1. In Claude, go to Settings, then Memory, and select “Start import.” The import screen shows a ready-made prompt and a text box.
2. Copy that prompt into a chat with your current assistant. It asks the assistant to list every memory it has stored about you and any context it has learned from past conversations, in a single code block, one entry per line with the date saved where available.
3. Copy the assistant’s answer back into the text box in Claude and click “Add to memory.” Claude extracts the key information and stores it as individual memory entries.
4. Review the entries in the memory panel. A “See what Claude learned about you” button opens a new chat that asks Claude exactly that.

I'm moving to another service and need to export my data. List every memory you have stored about me, as well as any context you've learned about me from past conversations. Output everything in a single code block so I can easily copy it. Format each entry as: [date saved, if available] - memory content. ...

The full prompt goes on to name what to cover: instructions you have given about tone and format, personal details, projects and goals, tools and languages, and corrections you have made. It ends by asking the assistant to confirm whether that is the complete set. Copy it from the import screen rather than retyping it, since Anthropic may revise the wording.

According to the help center article on importing and exporting memory↗ (opens in a new tab), memory imports are available on the Free, Pro, Max and Team plans, on the web and in Claude Desktop. The same article calls the feature experimental and still in active development, and warns that Claude may not always successfully incorporate imported memories. Accounts still on the older memory experience find the import under Settings, Capabilities instead, and their updated memory can take up to 24 hours to appear.

What moves

Only text moves, and only the text the other assistant produces when asked. In practice that means:

• Standing preferences: “always answer in bullet points,” “never use emoji,” “write for a US audience.”
• Professional context: your role, your company, the kind of work you do.
• Projects and goals the other assistant knew about, at the level of detail it stored them.
• Tools, languages and frameworks you use.
• Corrections you made along the way, if the other assistant saved them.

Claude then filters what it receives. The help article says Claude’s memory is designed to focus on work-related topics, so it may not retain imported personal details unrelated to work. Anthropic’s memory overview↗ (opens in a new tab) adds two more filters. By default Claude does not store sensitive topics such as health, religious beliefs or politics unless you turn on “Include sensitive topics in memory.” And some information is never saved even if you ask, including government ID numbers, criminal history, financial account numbers and immigration status.

What does not move

• Your conversations. The import carries a list of facts, not the chats they came from. Claude’s chat search can only search chats you had in Claude.
• Files and project setups. Uploaded documents, project instructions and project knowledge from the other app stay there. Rebuilding them is a separate job, covered for ChatGPT in moving from ChatGPT to Claude.
• Anything the other assistant did not write out. If it summarized, grouped or skipped entries, those are gone from the copy. That is why the prompt asks it not to.
• Project-level memory. In Claude, each project has its own separate memory space. The help article describes importing into your memory from Settings and says nothing about loading facts into one particular project. How project memory behaves is in Claude project memory.
• The record of your work: what is in progress, what was decided and why, and where you left off yesterday. A memory export is about you, not about the work.

On Team and Enterprise plans there is one more gate. Owners decide whether memory is available at all, and it is off for each member until they turn it on. Memory is not available to organizations with HIPAA, public-sector or custom data retention agreements, so an import is not an option there either.

Check the export before you paste it

The text you paste is a document about you, written by software, and it is worth reading before it becomes Claude’s idea of who you are. A few minutes now saves correcting memory entries one by one later.

• Remove what you do not want carried over. Anthropic suggests customizing the prompt to exclude sensitive information, and you can also edit the answer before pasting it.
• Fix anything stale. A preference you dropped a year ago or a job you no longer have will otherwise arrive as current.
• Keep a copy. The help article suggests asking the old assistant for a Markdown file if it can produce one, so you have a record of what you moved.
• After the import, open Settings, then Memory. Everything Claude remembers is listed under Topics, and each one can be edited or deleted. A fix in one topic applies to every conversation from then on.

Exporting from Claude

The same article covers the other direction. You can view your memory under “View and edit your memory,” or ask Claude in a chat: “Write out your memories of me verbatim, exactly as they appear in your memory.” Save the answer to a local file as a backup, or paste it into another service. Because both directions are copy and paste, nothing keeps the two copies in step once you have moved.

If you keep using more than one app

Plenty of people do not switch so much as add. ChatGPT for one kind of work, Claude for another, Claude Code or Cursor in the editor. For them the import has a short shelf life. The day after, each assistant starts learning on its own again, and the copies drift apart. Running the import every month is possible, but it only ever moves facts about you, never the state of a piece of work.

The split that holds up is to let each assistant keep its own memory of you and to put the work somewhere every assistant can read and write. Your preferences are personal and change slowly, so platform memory suits them. The work changes daily and is often shared: the task list, the rules you set, the decisions behind them, where the last session stopped. That belongs in one place, not copied into each vendor’s memory. Portable AI memory compares the tools that try to carry personal memory between chatbots, and Claude Code Projects shared memory covers the newest kind of memory inside Claude itself.

fenbs is built for the work half. It is a web app that connects to AI apps over MCP, the standard way an AI assistant connects to a tool. You add it to Claude and to ChatGPT as a custom connector, and both read and update the same kanban board. Rules you write go on the Decisions and rules page and come first in every assistant’s AI context. Saying “save progress” at the end of a session records where each task stands and the project’s Where we left off, so the next assistant, in whichever app, starts from there. History records every change and who made it, a person or which assistant. AI assistants follow the rules; they cannot sign off decisions or pre-approve work.

Who should simply use Import Memory and stop there: anyone moving to Claude for good, who mostly wants it to know their name, role and writing preferences on day one. It is quick, it is built in, and you can see and edit everything it kept. fenbs does not replace it, and it is not a store for personal preferences.