<br />

> [!NOTE]
> **Public preview:** The federated search and actions features are in public preview. This feature is subject to the "Pre-GA Offerings Terms" in the [General Service Terms section of the Service Specific Terms](https://cloud.google.com/terms/service-terms). Pre-GA features are available "as is" and might have limited support. For more information, see the [launch stage descriptions](https://cloud.google.com/products#product-launch-stages).

<br />

This page describes how to create a data store and connect Windsor.ai to
Gemini Enterprise.

## Before you begin

Before you set up your Windsor.ai connection, do the following:

1. Grant the Discovery Engine Editor role (`roles/discoveryengine.editor`). This role is required for the user to create the data store. To grant this role, do the following:
   1. In the Google Cloud console, go to the **IAM** page.

      [Go to IAM](https://console.cloud.google.com/iam-admin/iam)
   2. Locate the user account and click edit **Edit**.
   3. Grant the `Discovery Engine Editor` role to the user. For more information, see [IAM roles and permissions](https://cloud.google.com/gemini/enterprise/docs/access-control).


> [!NOTE]
> **Note:** Windsor.ai uses Google-managed OAuth. You don't need to create your own OAuth application or provide a client ID and client secret. You authorize the connector with your Windsor.ai account when you enable connector in app.

<br />

## Create the Windsor.ai data store

To create the Windsor.ai data store, follow these steps:

1. In the Google Cloud console, go to the Gemini Enterprise page.

   [Go to Gemini Enterprise](https://console.cloud.google.com/gen-app-builder/engines)
2. Select or
   [create](https://docs.cloud.google.com/resource-manager/docs/creating-managing-projects.md.txt)
   a Google Cloud project.

3. Click **Data stores** in the navigation menu.

4. Click **Create data store**.

5. In the **Third-party connectors** section, select
   **Windsor.ai**.

6. In the **Data** section:

   1.
      If the create data store flow shows an **Advanced options** section, you can select the **Enable Static IP Addresses** checkbox within it. This ensures the data store uses a fixed set of IP addresses for outbound traffic, suitable for allowlisting on the source system. For more information, see [Configure static IP egress](https://docs.cloud.google.com/gemini/enterprise/docs/connectors/configure-static-ip-egress.md.txt).

   2. Click **Continue**.

   3. In the **Entities to search** section, choose which entities
      Gemini Enterprise should scan. For Windsor.ai, select
      **Connectors**.

7. Click **Continue**.

8. In the **Configuration** section:

   1. From the **Multi-region** list, select the location for your
      data store.

   2. In the **Your data connector name** field, enter a name for
      your data store.

   3. Optional: In the **Tag** field, enter a stable identifier for
      this data store across versions. This value is immutable
      after you set it.

   4. If you selected `us` or `eu` as the location, configure the
      **Encryption settings**:

      - Optional: If you haven't configured single-region keys, click **Go to settings page** to do so. For more information, see [Register a single-region key for third-party connectors](https://docs.cloud.google.com/gemini/enterprise/docs/cmek.md.txt#third-party-connectors-key-registration).
      - Select **Google-managed encryption key** or **Cloud KMS key**.
      - If you selected **Cloud KMS key** :
        - In the **Key management type** list, select the appropriate type.
        - In the **Cloud KMS key** list, select the key.
   5. Optional: In the **Sensitive Data Protection policy**
      section, attach an existing
      [Sensitive Data Protection content policy](https://docs.cloud.google.com/sensitive-data-protection/docs/content-policy.md.txt)
      to inspect and block sensitive data that
      Gemini Enterprise retrieves through this data store. For more
      information, see
      [Protect sensitive data in sources](https://docs.cloud.google.com/gemini/enterprise/docs/protect-sensitive-data.md.txt).
      If your organization doesn't require a Sensitive Data Protection
      content policy, you can skip this step.

9. Click **Create**.

On the **Data Stores** page, click your data store name to see the status. After
the data store state changes from **Creating** to **Active**, the
Windsor.ai data store is ready to be used.

After creating the data store, connect the data store you just created to an
existing app or [create a new app](https://docs.cloud.google.com/gemini/enterprise/docs/create-app.md.txt) and
[connect](https://docs.cloud.google.com/gemini/enterprise/docs/connect-existing-data-store) it to the new app, and
[authorize](https://docs.cloud.google.com/gemini/enterprise/docs/connect-existing-data-store#user_authorization)
Gemini Enterprise to access Windsor.ai before executing any queries.

## Enable actions

By default, all actions are turned off. When you connect Windsor.ai, all
of its tools are imported and displayed as actions in the Gemini Enterprise
data store. To enable these actions for your users, follow these steps:

1. Go to your Windsor.ai data store.

2. Click **Actions \> Reload custom actions** to reauthenticate.

   > [!NOTE]
   > **Note:** This action performs a `tools/list` call on the MCP server to retrieve available tools, which are then displayed on the screen.

3. Select the actions to enable.

4. Click **Enable actions**.


## Data handling and query execution

This section describes how Gemini Enterprise manages your query and the privacy implications of using the federated data store.

### Query execution

After you authorize Windsor.ai and send a search query to Gemini Enterprise:

- Gemini Enterprise sends your search query directly to the Windsor.ai API.
- Gemini Enterprise blends the results with those from other connected data sources and displays a comprehensive search result.

### Data handling

When using third-party federated search, the following data handling rules apply:

- Your query string is sent to the third-party search backend (Windsor.ai API).
- These third parties may associate queries with your identity.
- If multiple federated search data sources are enabled, the query might be sent to all of them.
- After the data reaches the third-party system, it is governed by that system's terms of service and privacy policies.

> [!NOTE]
> **Note:** To improve accuracy, an LLM may rewrite your query before it is sent to Windsor.ai. This rewritten query can incorporate information from your session's query history. As a result, parts of your query history might also be included in the query sent to the Windsor.ai API.

<br />

## What's next

- To provide a user interface for querying your Windsor.ai data, [create an app](https://docs.cloud.google.com/gemini/enterprise/docs/create-app.md.txt) and [connect it to the Windsor.ai data store](https://docs.cloud.google.com/gemini/enterprise/docs/connect-existing-data-store).
- To view and manage the list of actions you enabled for the Windsor.ai data store, see [View and manage actions](https://docs.cloud.google.com/gemini/enterprise/docs/manage-actions).
- To preview how your search results appear after your app is set up, see [Get search results](https://docs.cloud.google.com/gemini/enterprise/docs/preview-search-results.md.txt).
- To enable alerts for the data store, see [Configure alerts for third-party data stores](https://docs.cloud.google.com/gemini/enterprise/docs/configure-periodic-data-alerts).