How to Move a Long AI Conversation to Another Assistant Without Losing Context

You're deep into a long conversation with one AI assistant and you want to take it somewhere else. Maybe you want a second opinion from a different model, or you want to continue in another tool, or you want another assistant to summarize what the first one actually said. Starting a new chat and retyping the background is slow, and you'll forget things. This guide shows three ways to carry a conversation across, with a handoff prompt you can copy.

This was my own reason for building ChatGPT Extractor. I kept wanting to take a long thread to another assistant, and copying it message by message wasn't realistic.

Why you'd move a conversation

• A second opinion. A different assistant may approach the same problem differently, or catch something the first one missed.
• Continuing elsewhere. You hit a usage limit, or you prefer another tool for the next stage of the work.
• Branching. You want to try a different direction without losing the original thread.
• Understanding. A long answer is hard to follow, and another assistant can summarize it in plain language.

A new conversation starts blank. Unless an assistant has its own memory feature, it knows only what you paste into it. So the real question is what to paste.

Three ways to hand a conversation over

Approach	Best for	Weak point
Paste the full transcript	Short or medium chats where exact wording matters	May not fit; the new assistant can latch onto mistakes made early on
Hand over a brief only	Long chats where you're changing direction	A summary is lossy and can drop details
Brief plus transcript	Important work	Takes longest to set up

If you're unsure, use the third option. The brief tells the new assistant what matters, and the transcript lets it check the details.

Step 1: Get the full conversation as text

1. Open the conversation and use the Share button to create a public link.
2. Copy the link and paste it into ChatGPT Extractor.
3. Choose Markdown as the format. It keeps headings, lists, and code blocks intact and reads cleanly as plain text.

The same steps work for shared links from Claude, Gemini, Grok, Meta AI, and Perplexity. If you'd rather start from the official account export instead, this guide explains why it's the wrong tool for one conversation.

A share link is public, so anyone with the link can open that conversation. Remove the link when you're done, and don't share chats that contain private details.

Step 2: Ask the first assistant to write the brief

The assistant that holds the full conversation is the best one to summarize it. Paste this into the original chat:

I'm about to continue this conversation in a different AI assistant that
has none of this context. Write a handoff brief I can paste into it.

Include:
1. The goal: what I'm trying to achieve, in two or three sentences.
2. Decisions made so far, and the reasons for them.
3. Constraints and preferences I've stated (tools, versions, tone,
   budget, anything I ruled out).
4. Current state: what's done, what's half-done, and any code, numbers,
   or wording that must be kept exactly. Quote those verbatim.
5. Open questions and the next step.

Keep it under 500 words. Don't add anything we didn't discuss. If
something is uncertain, mark it as uncertain.

Read the brief before you use it. If it left out something important, ask for that detail to be added.

Step 3: Start the new conversation

Open the new assistant and paste this, followed by the brief and the transcript:

Below is a handoff brief and the full transcript of an earlier
conversation with a different assistant. Read the brief first, then use
the transcript to check details. Continue from the "next step" in the
brief.

If the earlier assistant's reasoning looks wrong or incomplete, say so
instead of going along with it. Ask me before assuming anything the
transcript doesn't cover.

[BRIEF]

[TRANSCRIPT]

The line about disagreeing matters. Without it, a new assistant tends to treat the earlier conversation as settled and builds on top of it, mistakes included. That defeats the point of getting a second opinion.

Tips for a clean handoff

• Remove secrets first. Delete API keys, passwords, and personal details from anything you paste into another service.
• Keep the goal at the top. Put the reason for the conversation before the history, so the new assistant knows what it's helping with.
• Quote exact things exactly. Code, numbers, and wording you want preserved should appear verbatim, not paraphrased.
• Treat the summary as lossy. A brief is only as good as what it includes. For high-stakes work, compare it against the transcript before relying on it.
• Don't paste more than you need. If a transcript is very long, the brief alone may work better than a wall of text. If an assistant rejects or cuts off your paste, switch to the brief.
• Verify facts again. A second assistant can repeat the first one's errors with confidence. For anything that matters, check it against a reliable source.

Frequently asked questions

Will the new assistant remember my old chat? Not by itself. Unless it has its own memory feature, it only knows what you paste in. That's why the brief and transcript matter.

Which format should I export? Markdown works best for pasting into another assistant because it keeps structure without extra clutter. Plain text also works. JSON is better for scripts than for conversations with another AI.

Can I upload the file instead of pasting? If the assistant accepts file uploads, yes. If it doesn't, paste the text directly.

Is it better to paste the transcript or a summary? It depends on length and stakes. Short chats can go in whole. For long ones, use a brief, and add the transcript when details matter.

Also read: Why ChatGPT's Built-In Export Is All-or-Nothing • Export Formats Explained • "Private" AI Chats Turned Up on Google

More guides are on the blog, and release notes are on the changelog.