# Memory

Memory is what the AI has learned about **you**: your role, your preferences, how you
like things written. It personalizes your own chats and nothing else.

Open **Memory** from the menu under your name.

## What memory holds

Facts and preferences you would otherwise repeat in every conversation:

- *I work in marketing operations and report to the CMO*
- *Write in American English, never British*
- *Our fiscal year starts in April*
- *Prefer bullet points over paragraphs for summaries*

Add them yourself from this screen, or let them accumulate from what you tell the AI in
conversation.

## Manage memories

| | |
|---|---|
| **Pause** | Stop using a memory without deleting it. |
| **Resume** | Put it back into use. |
| **Delete** | Remove it. |
| The master toggle | Turn memory off entirely. Saved memories are kept but not used. |

Review your memories occasionally. Preferences that made sense during one project keep
applying afterward, and a stale memory is a quiet, persistent source of odd answers.

## Memory is private

Your memories are visible only to you. They are not used in your colleagues'
conversations and are not part of the workspace's knowledge.

That is the difference between the two features:

| | |
|---|---|
| **Memory** | About you. Private. Used only in your own chats. |
| [**Knowledge bases**](/guide/ai/knowledge-bases) | About your organization. Shared. Used by everyone's chats, and by widgets, decks and generated content. |

If you are adding a memory that everyone would benefit from (a fact about your
products, a policy, a name), put it in a knowledge base instead, where one correction
fixes it for the whole organization.

<Callout type="note">
	Do not put sensitive information in memory. It is a convenience for personalizing
	answers, and the AI reads it in every conversation you have.
</Callout>