# How do I move my AI memory and context to Gryz?

Source: https://www.gryz.ai/gryzle/move-your-ai-memory-to-gryz/
Updated: 2026-11-19

**Short answer:** Connect Gryz to the AI you use today, then ask it to save what it knows about you to Gryz memory and publish your important work as Gryz documents. After that, ChatGPT, Claude, Cursor, Codex and any other connected agent all start from the same context. You're no longer tied to the one that happened to learn it first.

Facts about ChatGPT, Claude and Codex below are from their own help pages and setup guides, as of November 2026.

## You're more locked in than you think [#youre-more-locked-in-than-you-think]

You didn't pick your AI for life. You picked one, used it for months, and it slowly learned:

* who you are and how you like work delivered
* your projects, people and shorthand
* the decisions you made and the ideas you ruled out
* the plans, specs and drafts you wrote together

None of that lives in the model. It lives inside one product. Try a different agent and it knows nothing, so you either start over or stay put.

That's the real lock-in. Not the model, the context.

## The fix: one place every agent reads [#the-fix-one-place-every-agent-reads]

Gryz is where your context lives outside any one AI:

* **Memory.** Small facts and decisions, each one showing which agent saved it.
* **Documents.** Plans, specs, reports and artifacts, with versions and comments.
* **Notes and tasks.** Rough thinking and next steps.

Any agent you connect to Gryz can read and write all of it. You can see, edit, export or delete everything in your dashboard.

The nice part: you don't have to copy anything by hand. The AI you use today can do the moving for you.

## Step 1: connect Gryz [#step-1-connect-gryz]

Connect Gryz to the AI that knows you best. Sign in once, no key to paste.

* **ChatGPT:** install [Gryz in ChatGPT](https://chatgpt.com/plugins/plugin_asdk_app_6acd840c118c81919b71f8a5783f8c96&#x29;, then type &#x2A;*@** and pick Gryz in a chat.
* **Claude:*&#x2A; Customize, then Connectors, **+ Add**, **Add custom connector**, and paste `https://www.gryz.ai/api/mcp`.
* **Claude Code, Cursor, Codex and others:** see [MCP setup](/docs/mcp/setup/).

Use the **same Gryz account** everywhere. That's what makes it shared.

## Step 2: move your memory [#step-2-move-your-memory]

### From ChatGPT or Claude [#from-chatgpt-or-claude]

Paste this into a chat with Gryz connected:

```text
Go through everything you remember about me: who I am, my work, projects, people, preferences and decisions. Save each lasting fact to Gryz memory, one fact per item. Skip anything that's out of date. When you're done, list what you saved.
```

What to expect:

* **It moves what it can see.** That's its saved memory, your custom instructions and whatever it can pull from past chats. It can't read every old conversation word for word.
* **It may ask before each save.** Approve them, or ask it to start with the 20 most important facts.
* **Want to check the list first?** ChatGPT shows its saved memories in Settings, under Personalization, then Memory ([Memory in ChatGPT](https://help.openai.com/en/articles/8590148-memory-in-chatgpt)). Claude can export its memory too, a feature Anthropic calls experimental ([Import and export your memory](https://support.claude.com/en/articles/12123587-import-and-export-your-memory-from-claude)). Paste either list into the chat and ask it to save the good ones.

### From Claude Code or Codex [#from-claude-code-or-codex]

Coding agents are the easiest case. Their memory is plain files on your computer, so nothing gets missed.

* **Claude Code** keeps instructions in `CLAUDE.md` files and its own notes in a memory folder for each project.
* **Codex** reads `AGENTS.md` files.

Ask:

```text
Read my CLAUDE.md, AGENTS.md and any memory files for this project. Save each lasting fact, convention and decision to Gryz memory for this project. Don't copy secrets, keys or passwords. List what you saved.
```

Now Cursor, ChatGPT or a new teammate's agent knows the same conventions without anyone rewriting them.

## Step 3: move your documents and artifacts [#step-3-move-your-documents-and-artifacts]

Memory is the small stuff. Your real work is in the documents.

### Artifacts in ChatGPT or Claude [#artifacts-in-chatgpt-or-claude]

Open the chat with the artifact or document you want to keep and ask:

```text
Publish this to Gryz as a restricted document.
```

Gryz keeps Markdown and HTML, so a plan, a report or an interactive page all work. It's private until you share it ([Visibility and sharing](/docs/publishing/visibility-and-sharing/)).

Go one chat at a time. Asking for "all my artifacts" won't work well, because the AI can only reach what's in the chat you're in, plus whatever chat search turns up.

### Files in a repo [#files-in-a-repo]

Ask your coding agent:

```text
Publish the plans and specs in docs/ to Gryz as restricted documents, one per file.
```

A free account keeps up to 3 private documents, so start with the ones that matter most. Now those docs can be read, commented on and edited from any agent, not just the one with your repo open.

## Step 4: tidy up [#step-4-tidy-up]

Give it five minutes:

* Open **Memory** in Gryz and delete anything stale or wrong. Agents save a mix of good facts and old ones. See the [Agent Memory overview](/docs/agent-memory/overview/).
* Open your documents and share the ones your team should see.
* Add one line to each AI's own instructions so it checks Gryz on its own:

```text
When I ask about my work, plans, projects or past decisions, check Gryz first. When I decide something or tell you a preference, save it to Gryz.
```

## Test it [#test-it]

Open a different agent than the one you moved from, connect Gryz, and ask:

```text
What do you know about me and my current projects? Check Gryz.
```

If it answers like it's known you for months, you're free.

## Limits worth knowing [#limits-worth-knowing]

* **It's a copy, not a sync.** Your old AI keeps its own memory. From now on, what matters is what gets saved to Gryz.
* **Chats don't move.** Only facts and documents your agent saves on purpose.
* **Free has limits.** A free Gryz account keeps 50 memories and 3 private documents. A long memory list may need trimming. See [pricing](/pricing/).
* **Gryz doesn't run your chats.** Your agents still do the work. Gryz holds what they all need to know.

More on these topics: [Agent memory](/topics/agent-memory/) and [Shared context for AI agents](/topics/shared-context/).

## FAQ [#faq]

**Can I transfer my ChatGPT memory to Claude or Codex?**
Yes, through Gryz. Move it from ChatGPT into Gryz once, then connect Claude or Codex to the same Gryz account and they read it too. Claude also has its own one-time import, but anything saved to Gryz later shows up in every connected agent. More on using both: [Share context between ChatGPT and Claude](/gryzle/share-context-between-chatgpt-and-claude/).

**Can I just tell ChatGPT to move everything to Gryz?**
Yes, once Gryz is connected. It moves what it can see: its saved memories, your instructions and what it finds in past chats. Use the prompt in step 2.

**Can Claude move my artifacts?**
Yes, one chat at a time. Open the chat and ask Claude to publish the artifact to Gryz.

**Will my old AI lose its memory?**
No. Nothing is deleted. Gryz gets a copy.

**Which agent is best for moving?**
Coding agents like Claude Code and Codex, because their memory is files they can read in full. Chat apps move what they can see.

**Who can see what I move?**
Only you and the agents you connect. Memories are private to your account, and documents are restricted until you share them. Gryz does not use your content to train AI models. Each AI you connect follows its own provider's terms.

**Is Gryz free?**
Yes. A free account covers everything on this page, within its limits. See [pricing](/pricing/).

Use the best AI for each job and switch as often as it pays to. What they know about you stays in one place you own.
