Add Azure AI Search as a knowledge source for agents in Copilot Studio.

Configure Azure AI Search

To complete the connection, you need an Azure account. If you don't already have an Azure account, you can create one at Microsoft Azure .

Here, key-based authentication is used to connect to Azure AI Search.

Copilot Studio supports vectorized indexes using built-in vectorization. Prepare your data and select an embedded model, then use Import and vectorize data in Azure AI Search to create a vector index. This method allows the system to use the same embedded model used to digitize the data, and also digitize the incoming prompt at runtime, minimizing the need to write special functions to perform the same task.

Copilot Studio also supports semantic ranking. This feature also needs to be configured in Azure AI Search before adding the connection in Copilot Studio.

Establish a connection to Azure AI Search

Note :

1. You must add Azure AI Search via the official data connection. Do not manually configure the endpoint and API key.
2. If you create an unsupported Azure AI Search connection, Copilot Studio may cache a failed data connection, preventing the Azure AI Search connection dialog from loading. When this happens, the dialog may appear briefly before reporting an error. Currently, Copilot Studio does not provide an interface to delete a failed Azure AI Search data connection. Because data connections are managed at the environment level, this issue can affect all agents.
3. To restore the connection, reset the agent's external access or delete and recreate the affected agent. When re-adding Azure AI Search, use Data sources → Azure AI Search with Entra ID authentication , not the API key.

1. Open the agent.

2. Select Add knowledge from the Overview or Knowledge page , or Properties of the Generative Answers node.

3. From the Add knowledge dialog box , select Featured .

4. Select Azure AI Search .

5. Select Create new connection .

6. Select Authentication type . You have the following options:

1. Access Key
2. Client Certificate Auth
3. Service principal (Microsoft Entra ID application)
4. Microsoft Entra ID Integrated

7. Enter the necessary details for the authentication type. For key-based authentication as described in the previous configuration section, enter the Azure AI Search Endpoint URL and Azure AI Search Admin Key .

8. Select Create to complete the connection. A green checkmark will appear to confirm the connection.

9. Select Next .

10. Import the Azure AI Search vector index to be used. Only one vector index can be added.

11. Select Add to agent to complete the connection.

After you add the connection, it will appear in the knowledge source table. The status displayed is Progress while Copilot Studio indexes the metadata in the tables. Once the indexing process is complete, the status will be updated to Ready , and you can begin testing the knowledge source. During testing, you can verify that the correct references have been called by reviewing the files and citations cited by the agent.

Return the quote

To return citations when using Azure AI Search in Copilot Studio, include a URL field with the actual link to the document in the search index. When the field metadata_storage_pathis included in the index, Copilot Studio will interpret it as a citation. However, if the field does not exist, Copilot Studio will treat any field containing a full URL link as a citation.

Note : When configuring citations in Azure AI Search, ensure that users in your agent have the necessary permissions to access the data sources that the citations point to. For example, if you add a URL to the search index that points to a website or knowledge base, users will have access to those sources. If the URL points to a restricted data source, users will not be able to access the generated citations.

Virtual network support

Copilot Studio supports Azure AI Search indexes configured for a virtual network. This configuration allows you to use private endpoints to securely connect your search indexes.

In Copilot Studio, follow the steps above to complete the connection configuration.