Importing ChatGPT and Claude memory
Bring what other assistants already know into Memtro.
Neither ChatGPT nor Claude exposes its memory through an API, so Dashboard → Import works from what they do offer officially: the memory summary each product shows in its settings, and the data export.
| Source | Memories | Conversation history |
|---|---|---|
| ChatGPT | Settings → Personalization → Memory → Manage memories: copy the list, or run the extraction prompt shown on the Import page | Settings → Data controls → Export data → upload conversations.json |
| Claude | Settings → Memory: copy the summary, or run the extraction prompt | Settings → Privacy → Export data → upload conversations.json |
Pasting memories
Paste the lines into the Import page and choose a scope (personal or shared). Each line is stored verbatim, deduplicated by similarity, and then passed once through entity extraction so the people and projects in them join the graph.
Uploading an export
Export files are parsed in your browser, so only compact transcripts are uploaded. Conversations go through the same extraction pipeline the chat endpoint uses, in resumable batches: a multi-thousand-conversation history survives interruptions and can be paused and resumed. Re-importing the same export skips conversations already processed.
After importing
Open Dashboard → Memory to see what came in, with source badges, and delete anything you do not want kept. Memtro's scope rules apply: facts about you stay personal unless you imported into the shared scope on purpose.