> ## Documentation Index
> Fetch the complete documentation index at: https://help.statisfy.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Managing Datasets

> Create, configure, and organize Knowledge Base datasets.

Datasets are collections that organize your documents into logical groups. Each dataset can be configured independently with its own Stella search settings and connected sources.

### Create a Dataset

<Note>
  Creating datasets requires Admin access.
</Note>

1. Navigate to **Knowledge Base** from the main menu

2. Click **Add Dataset**

3. Enter the dataset details:
   * **Name**: A unique name for the dataset (required, max 100 characters)
   * **Description**: Optional details about the dataset's purpose

4. Configure settings:
   * **Searchable in Stella**: Toggle on to enable AI-powered search

5. Click **Save**

### Edit a Dataset

1. Find the dataset you want to edit

2. Click the **Settings** (gear) icon on the dataset card

3. Update the dataset properties:
   * Change the name or description
   * Toggle Stella searchability
   * Connect or disconnect external sources

4. Click **Save** to apply changes

### Enable Stella Search

When "Searchable in Stella" is enabled:

* All documents in the dataset are indexed for semantic search
* Stella AI agents can query the dataset when answering questions
* Results are ranked by relevance, not just keyword matches

**To enable:**

1. Edit the dataset settings
2. Toggle **Searchable in Stella** on
3. Save changes

<Tip>
  Enable Stella search on datasets containing documentation, FAQs, or reference materials that AI agents should use when responding to queries.
</Tip>

### Connect External Sources

Datasets can automatically import documents from external sources:

**Currently Supported:**

* **Webpage** — enter a website URL and Statisfy crawls the site (visits its pages and copies their content) into the dataset
* **Google Drive** — sync specific Drive folder or file URLs into the dataset. See the [Google Drive integration guide](/integrations/google_drive_integration) for supported URL formats and account-level options
* **Intercom** — import articles from your Intercom Help Centers

**To connect a source:**

1. Edit the dataset settings
2. Click **Connect Source**
3. Select the source type:
   * **Webpage**: enter the URL of the site you want to import
   * **Google Drive**: add one or more Google Drive folder or file URLs; each URL is synced into the dataset. You can also point the source at an account-level custom field that contains Google Drive URLs, so each account syncs its own content
   * **Intercom**: choose the Help Center to sync
4. Save to start the import

<Note>
  The Google Drive and Intercom options require the corresponding integration to be connected first. If a source appears disabled, connect it under **Integrations** and try again.
</Note>

**Keeping sources up to date:**

* Content from connected sources is refreshed automatically every day, so your dataset stays in sync without manual action
* To pull in changes right away, click the **refresh** icon on the connected source card in the dataset settings, or trigger a refresh from the dataset's detail page — you'll see a "Refresh triggered" confirmation

### Delete a Dataset

<Warning>
  Deleting a dataset permanently removes all documents within it. This action cannot be undone.
</Warning>

1. Find the dataset you want to delete

2. Click the **three-dot menu** on the dataset card

3. Select **Delete**

4. Confirm the deletion

**Restrictions:**

* Datasets assigned to specific AI agents cannot be deleted
* You must first remove the agent assignment before deleting

### Dataset Best Practices

<Accordion title="Organize by topic or use case">
  Create separate datasets for different knowledge domains:

  * Product documentation
  * Support FAQs
  * Onboarding materials
  * Internal processes

  This makes it easier to control which AI agents have access to which content.
</Accordion>

<Accordion title="Use descriptive names">
  Choose names that clearly indicate the dataset's contents:

  * "Product Documentation" instead of "Docs"
  * "Customer Onboarding Guide" instead of "Guide"
</Accordion>

<Accordion title="Review before enabling Stella search">
  Before enabling Stella search, verify that:

  * Documents are appropriate for AI responses
  * Content is accurate and up-to-date
  * No sensitive internal information is included
</Accordion>