# Creating Quick Sight Topics
<a name="topics-create"></a>


|  | 
| --- |
|  Applies to:  Enterprise Edition  | 


|  | 
| --- |
|    Intended audience:  Amazon Quick administrators and authors  | 

To create a Topic, you add one or more enriched datasets, define relationships between them, and optionally add custom instructions that guide how the AI interprets cross-dataset queries.

## Creating a Topic
<a name="topics-create-how"></a>

Use the following procedure to create a Topic.

**To create a Topic**

1. On the Quick homepage, choose **Data** in the navigation pane at left.

1. Choose the **Topics** tab, and then choose **Create topic**.

1. Enter a topic name (for example, "Retail Sales Analytics") and an optional description, then choose **Create**.

1. Choose **Add dataset** and select the datasets you want to include in the Topic (up to 12 datasets).

1. After adding datasets, define the relationships between them. For more information, see [Defining relationships between datasets in a Topic](topics-relationships.md).

1. (Optional) Add custom instructions that provide cross-dataset business rules. For more information, see [Adding custom instructions to a Topic](topics-custom-instructions.md).

1. Choose **Publish** to make the Topic available to users.

After publishing, share the Topic with business users so they can ask natural language questions or use it in analysis. For more information, see [Sharing Quick Sight Topics](topics-sharing.md).

## Prerequisites
<a name="topics-create-prerequisites"></a>

Before creating a Topic, confirm that you have the following:
+ Amazon Quick Enterprise Edition enabled with Author or Admin role.
+ One or more datasets enriched with semantic metadata (column descriptions, synonyms, and semantic types). For more information, see [Data Preparation Experience (New)](data-prep-experience-new.md).
+ Datasets representing a dimensional model (fact tables and dimension tables) loaded into SPICE or accessible through a supported Direct Query source.
+ Permissions to create Topics and manage datasets.