Creating and Managing Knowledge Stores

Organizing Knowledge Stores

The way you organize Knowledge Stores directly affects retrieval accuracy, maintainability, and reuse. Because a single Knowledge Store may be referenced by multiple agents, you should clearly define its scope and content. Use the following best practices when creating Knowledge Stores:

• Create a separate Knowledge Store for each distinct use case or business function. Clear boundaries between Knowledge Stores make it easier to manage content and ensure that agents retrieve only relevant information.

• Keep data domains separate. Avoid combining unrelated data into a single Knowledge Store, as this can introduce noise into retrieval and reduce the accuracy of responses.

In general, prefer smaller, well-scoped Knowledge Stores. It is easier to combine Knowledge Stores later than to separate data after it has been grouped together.

Understanding the Knowledge Store Interface

Navigate to Management > AI Assistants >Knowledge Stores, and you will see a table of your existing Knowledge Stores. The table has three tabs, System, containing Knowledge Stores that the Data Studio has automatically created to support built-in features, SQL, containing Knowledge Stores designed for structured data from tables, and Document, containing Knowledge Stores designed for unstructured data. The details contained in each of the tabs of the table are as follows:

You can expand a Knowledge Store by clicking the + icon to view the following details about the Narratives associated with that Knowledge Store:

Creating a Knowledge Store

When creating a Knowledge Store, you are defining a new collection of information that assistants can reference.

1. Click the Management icon in the application menu.

2. Under AI Assistants, click Knowledge Stores.

3. Navigate to the tab for the data type you want to make a new Knowledge Store for.

4. Click + New SQL Store or + New Document Store.

5. Provide the following information on the Create Knowledge Store screen. Complete the following information:

   Knowledge Store Name

   Required— The unique name used to identify this Knowledge Store. It should be descriptive and concise.

   Knowledge Store Type

   Required—The type of Knowledge Store. SQL Based (Collection) or Document Based (Document Collection).

   Enabled

   Optional— Selecting this checkbox indicates that this Knowledge Store is active and available for users.

   User Roles

   Optional— Select one or more roles allowed to interact with this Knowledge Store.

   Description

   Required—Describe the contents, scope, and intended use of the Knowledge Store. This description helps the Assistant determine when the Knowledge Store is relevant and how to use its data during retrieval. For best results, include the type of data it contains, the kinds of questions it should answer, and any important scope boundaries or limitations.

   Embedding Model

   Required—The embedding model used for processing and storing data in this Knowledge Store. Choose an embedding model based on your retrieval needs:

   • text-embedding-3-small — Best for general vector search at scale. Faster and lower cost, with slightly lower precision.

   • text-embedding-3-large — Use for complex or high-stakes retrieval where meaning is subtle or requires inference. Higher accuracy and greater resource requirements.

   • text-embedding-ada-002 — Legacy option. Use only for backward compatibility with existing Ada embeddings.

6. Once all the fields are completed, click Submit. The new Knowledge Store will appear on your Knowledge Store table.

The next step in this process is to Add Data to Your Knowledge Store.

Editing a Knowledge Store

1. Click the Management icon in the application menu.

2. Under AI Assistants, click Knowledge Stores.

3. Click Edit Store.

4. Modify the fields you would like to change.

5. Click Submit.

Changes take effect for future data processing. Existing indexed data may need to be reprocessed depending on the fields modified.

Deleting a Knowledge Store

A Knowledge Store with the type Document Collection or Collection that has no Narratives associated with it can be deleted. System Knowledge Stores cannot be deleted. To do so:

1. Click the Management icon in the application menu.

2. Under AI Assistants, click Knowledge Stores.

3. Click the Edit menu.

4. From the menu, click Delete Store. A confirmation box appears.

5. Click Delete.

Deleted Knowledge Stores cannot be recovered.

Disabling or Enabling a Knowledge Store

1. Click the Management icon in the application menu.

2. Under AI Assistants, click Knowledge Stores.

3. Click the Edit menu.

4. From the menu, click Disable to disable the Knowledge Store or Enable to re-enable it.