> For the complete documentation index, see [llms.txt](https://docs.maiagent.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.maiagent.ai/maiagent-user-guide/en/work/maigpt/ai-memory.md).

# AI Memory

View and manage the preferences and facts your AI assistants have learned from conversations

The AI Memory page lets you view and manage the preferences and facts currently saved to your account. You can add, edit, delete, or reactivate your memories here.

{% hint style="info" %}
This item appears on the account page only when your organization has enabled AI Memory.
{% endhint %}

## When to use it <a href="#use-case" id="use-case"></a>

For example, a project manager wants to organize the communication preferences saved to her account. She can open <mark style="color:blue;">AI Memory</mark> to review existing entries, then manually add a preference to “reply in Traditional Chinese.” If her needs change, she can edit or delete it at any time.

## Open the memory list <a href="#open-memory-list" id="open-memory-list"></a>

Open the account menu in the upper-right corner and select <mark style="color:blue;">Profile</mark>. In the organization section, select <mark style="color:blue;">AI Memory</mark>. The page lists your memories. If there are none, it displays “No memories yet.”

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-abad881f3ab1ac20097bc14dacc629a5ff640bf4%2Fai-memory-list.png?alt=media" alt="Empty state of the AI Memory page, with the Add button in the upper-right corner"><figcaption><p>Empty state of the AI Memory list</p></figcaption></figure>

## Add a memory manually <a href="#add-memory" id="add-memory"></a>

{% stepper %}
{% step %}
On the <mark style="color:blue;">AI Memory</mark> page, select <mark style="color:blue;">Add</mark> in the upper-right corner.
{% endstep %}

{% step %}
Choose a type and enter a topic, summary, and details. Use English `snake_case` for the topic, such as `communication_style`. You can put each detail on a separate line.

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-89edbfeec742e3fab67138f346e8bb7c09d254dd%2Fai-memory-add.png?alt=media" alt="Add AI Memory form with type, topic, summary, and details fields"><figcaption><p>Fields required when adding a memory manually</p></figcaption></figure>
{% endstep %}

{% step %}
Select <mark style="color:blue;">Save</mark>. The new memory appears in the list.
{% endstep %}
{% endstepper %}

## Ask the assistant to remember in a conversation <a href="#save-in-chat" id="save-in-chat"></a>

Besides adding a memory manually on the <mark style="color:blue;">AI Memory</mark> page, you can tell the assistant “Remember…” in a conversation. The assistant turns the information into a memory and saves it to your memory list for use in later conversations.

The assistant may also save facts it considers useful over the long term, such as your job role, preferred response language, or frequently used skills.

### Requirements <a href="#save-in-chat-requirements" id="save-in-chat-requirements"></a>

The assistant can write a memory during a conversation only when all of these conditions are met:

* The organization has enabled AI Memory.
* <mark style="color:blue;">Vector Memory</mark> is enabled in the AI assistant's settings.
* The person in the conversation is an organization member. Conversations with anonymous visitors do not create memories.

{% hint style="warning" %}
The assistant does not save sensitive personal information such as credit card numbers, national ID numbers, passport numbers, bank account numbers, mobile phone numbers, email addresses, passwords, or API keys. If content contains such information, it is not saved as a memory, and the assistant explains that it cannot save it.
{% endhint %}

### Example <a href="#save-in-chat-scenario" id="save-in-chat-scenario"></a>

For example, a marketing planner is preparing a presentation for a new product in MaiGPT and wants to avoid repeating its code name in every conversation. She types, “Remember: our demo product's code name is Aurora. Use that code name whenever you write product descriptions.” After using <mark style="color:blue;">Save Memory</mark>, the assistant confirms that it has remembered. She can then see the memory on the <mark style="color:blue;">AI Memory</mark> page, labeled <mark style="color:blue;">User Request</mark>. When she starts another conversation and asks for a product description, the assistant uses the code name Aurora.

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-a97fd05d5498ebff8adf57d180d7d53c0e8603dd%2Fmaigpt-memory-save-in-chat.png?alt=media" alt="In a MaiGPT conversation, the user asks the assistant to remember the demo product code name Aurora; the steps show Save Memory and the assistant confirms"><figcaption><p>Ask the assistant to remember during a conversation; the steps show Save Memory</p></figcaption></figure>

{% hint style="info" %}
If a memory written by the assistant has the same topic as an existing memory, it updates that entry instead of creating a duplicate.
{% endhint %}

## Memory source labels <a href="#source-tags" id="source-tags"></a>

Every memory in the list shows its source. Memories written by the assistant during a conversation display one of these labels:

| Label                                                         | Meaning                                                                      |
| ------------------------------------------------------------- | ---------------------------------------------------------------------------- |
| <mark style="color:blue;">User Request</mark> (green)         | Content you explicitly asked the assistant to remember during a conversation |
| <mark style="color:blue;">Assistant Initiated</mark> (purple) | Content the assistant decided to save during a conversation                  |

The list may also show other source labels, such as <mark style="color:blue;">Added Manually</mark> for memories you added on this page, or <mark style="color:blue;">AI Extraction</mark> for information automatically extracted from conversations by the system. If you do not want to keep a memory, edit or delete it as described below.

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-1d67cfaf5ece4c4e3090598fce4d82f9d1d12ba6%2Fmaigpt-memory-source-tags.png?alt=media" alt="Two memories in the AI Memory list: one labeled Fact and AI Extraction, and another with topic demo_product_code labeled Instruction and User Request"><figcaption><p>Source labels in the memory list: AI Extraction and User Request</p></figcaption></figure>

## Manage existing memories <a href="#manage" id="manage"></a>

In the memory list, you can:

* Edit the topic, summary, and details.
* Delete memories that are no longer accurate or that you do not want the AI to use. Deleted memories cannot be restored.
* Reactivate expired entries in the “Expired Memories” section.
* Select Refresh to show the latest list.

{% hint style="warning" %}
Memories may affect answers in later conversations. Correct or delete information that is outdated, inaccurate, or should not be kept long term.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.maiagent.ai/maiagent-user-guide/en/work/maigpt/ai-memory.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.