How to Export Your ChatGPT History (and Actually Use It)

If you've used ChatGPT for more than a few months, there's a real archive in there: research you did once and never wrote down, decisions you talked through, drafts, plans, half-finished ideas. All of it lives in one app, findable only by scrolling, and none of it is available to any other AI you use.

OpenAI gives you a way to export all of it — every conversation, verbatim, in one file. This post walks through the export step-by-step, and then through the part most guides skip: doing something useful with the file. You'll end with your full history backed up in a place you control, searchable by meaning, from ChatGPT itself, Claude, Cursor, or any other AI tool you use.

Why export your ChatGPT history at all

• It's a backup.Your history exists in exactly one place, governed by one account. Accidentally delete a conversation, lose account access, or hit a policy change, and it's gone. An export is your copy.
• It's portable context.Everything you've ever explained to ChatGPT — your projects, preferences, past research — is context that Claude, Cursor, and every other AI you use starts without. Your history is the richest prompt you own.
• Scrolling is not search.ChatGPT's built-in search matches keywords in titles. Finding “that conversation where I compared database options in March” needs semantic search over the actual transcripts.

Step 1: Request the export from ChatGPT

1. In ChatGPT, open Settings → Data Controls → Export data and confirm.
2. OpenAI emails you a download link — sometimes within minutes, sometimes a day or two (OpenAI's own notice says it can take a few days, and larger accounts take longer). It's worth requesting it early.
3. The link expires 24 hours after it arrives, so download the archive when the email lands.

Export is available on Free, Plus, Pro, and eligible Edu plans. It isn't available inside Business/Enterprise workspaces — those are admin-controlled.

Step 2: Find conversations.json in the archive

The download is a zip file. Unzip it and you'll find a handful of files — the one that matters is conversations.json. That single file is your entire ChatGPT history: every conversation, every message, with titles and timestamps. (The archive also contains your images and account metadata; for text history, conversations.json is the whole story.)

You could stop here and file the zip away as a cold backup. But a JSON file you never open isn't much of an archive — the interesting step is making it usable.

Step 3: Load it into Engram

Engram is a memory service for AI tools: it stores conversations verbatim and makes them searchable by meaning from any connected AI. Importing your ChatGPT export takes a few minutes:

1. Create a free account and open your dashboard.
2. Drop conversations.json into the Import your history section. The file is parsed in your browser and imported with a live progress bar — no terminal required.
3. Prefer the command line? The CLI does the same thing:

npx @getengram/cli import conversations.json --dry-run  # preview what will import
npx @getengram/cli import conversations.json            # import for real

Every conversation is stored verbatim — no summarizing, no lossy “memory extraction” — then chunked and embedded so it can be searched by meaning, not just keywords. The free plan holds 10,000 messages permanently; if your history is bigger than that, the Pro plan holds a million.

What you can do once it's imported

This is where the export stops being a backup and starts being useful. Connect Engram to your AI tools and your history follows you:

• In ChatGPT— install Engram from the Plugin directory and ask: “Search my memory for the database comparison I did in March.” It finds the actual conversation, not a title match.
• In Claude or Claude Code— add Engram as an MCP server and your years of ChatGPT context are available in a completely different AI. Ask “what do you remember about my side project?” and it answers from conversations that happened somewhere else.
• In Cursor, Windsurf, VS Code — your coding agent can pull up the architecture discussion you had with ChatGPT last quarter while it edits your code today.

Setup guides for each tool are in the docs — most are one config block.

Keeping it current

The import covers everything up to the day you exported. Going forward, you don't re-export: with Engram connected, saving is part of the conversation — say “save this to Engram” in ChatGPT, or let your tools store context as they work. The imported archive and the new memories live in the same place and search as one history.

Common questions

Is my history private once it's in Engram?

Your conversations are stored verbatim in your account, searchable only with your API key. There are per-conversation privacy controls, and you can read the security docs for details. And it stays portable: one click in the dashboard exports everything back out as JSON, and deleting is always available.

How long does the import take?

A few minutes for most accounts. The dashboard importer shows live progress; the CLI's --dry-run flag previews counts before anything is written.

What about my images and files?

The importer reads the text of your conversations from conversations.json. Images and uploaded files in the archive aren't imported — keep the zip if those matter to you.

Does this work for Claude exports too?

Yes — the same importer accepts Claude's export format, so you can consolidate both histories into one searchable memory.

Getting started

Request your export now — OpenAI can take anywhere from minutes to a couple of days to send the link, so the sooner you ask, the better. Meanwhile, create a free Engram account; it's useful before the import even lands. When conversations.json is in hand, the step-by-step import guide takes it from there.