Skip to Content

Google Gemini connection in Jitterbit Studio

Introduction

A Google Gemini connection, created using the Google Gemini connector, establishes access to Google Gemini. Once a connection is configured, you can create instances of Google Gemini activities associated with that connection to be used either as sources (to provide data in an operation) or as targets (to consume data in an operation).

Create or edit a Google Gemini connection

A new Google Gemini connection is created using the Google Gemini connector from one of these locations:

An existing Google Gemini connection can be edited from these locations:

Configure a Google Gemini connection

Each user interface element of the Google Gemini connection configuration screen is described below.

Tip

Fields with a variable icon support using global variables, project variables, and Jitterbit variables. Begin either by typing an open square bracket [ into the field or by clicking the variable icon to display a list of the existing variables to choose from.

  • Connection name: Enter a name to use to identify the connection. The name must be unique for each Google Gemini connection and must not contain forward slashes / or colons :. This name is also used to identify the Google Gemini endpoint, which refers to both a specific connection and its activities.

  • Platform: Select the Google Gemini platform to connect to, either AI Studio or Vertex AI.

    • AI Studio: Select this option to connect to Google AI Studio. When this option is selected, this field is visible:

    • Vertex AI: Select this option to connect to the Gemini Enterprise Agent Platform (previously known as Vertex AI). When this option is selected, these fields are visible:

      • Project ID: Enter the project_id provided by the service account credentials JSON file.

      • Location: Enter the Google Cloud region where your Gemini Enterprise Agent Platform resources are hosted (for example, us-central1).

      • Client email: Enter the client_email provided by the service account credentials JSON file.

      • Private key string: Enter the private_key provided by the service account credentials JSON file. The private key should be entered in this format:

        -----BEGIN PRIVATE KEY-----
<private_key>
-----END PRIVATE KEY-----

        Note

        All newline characters (\n) present in the private_key string must be replaced with a carriage return in the Private key string field. Otherwise, an error will occur when testing the connection.

  • Store chat context across operations: Click to enable short-term memory that stores chat context beyond the scope of a single operation.

    Important

    Enabling this setting stores chat context across operation chains for cloud agent groups only. Private agents always store chat context in memory for operations within the same operation chain that share the same chatId, regardless of this setting.

    When enabled, these additional settings are shown:

    • Chat context TTL: Enter the length of time to retain chat context in the short-term memory in seconds. The default value is 86400 seconds (1 day).

    • Maximum messages saved in context: Enter the maximum number of messages to save to short-term memory's chat context. The default and minimum value is 10 messages.

    • When chat context is not found: Select the action to take if chat context isn't found in the short-term memory:

      • Create a new context: (Default.) A new chat context is created.

      • Fail the request: The request fails.

  • Optional settings: Click to expand additional optional settings:

    • Timeout: Enter the timeout value in seconds.

    • Use Proxy Settings (Private Agent Only): When using a private agent, this setting can be selected to use private agent proxy settings.

  • Test: Click to verify the connection using the specified configuration. When the connection is tested, the latest version of the connector is downloaded by the agent(s) in the agent group associated with the current environment. This connector supports suspending the download of the latest connector version by using the Disable Auto Connector Update organization policy.

  • Save: Click to save and close the connection configuration.

  • Discard changes: After making changes to a new or existing configuration, click to close the configuration without saving. A message asks you to confirm that you want to discard changes.

  • Delete: After opening an existing connection configuration, click to permanently delete the connection from the project and close the configuration (see Component dependencies, deletion, and removal). A message asks you to confirm that you want to delete the connection.

Next steps

After a Google Gemini connection has been created, you place an activity type on the design canvas to create activity instances to be used either as sources (to provide data in an operation) or as targets (to consume data in an operation).

Menu actions for a connection and its activity types are accessible from the project pane and design component palette. For details, see Actions menus in Connector Basics.

These activity types are available:

  • Register Tools: Registers functions (tools) to the specified model in Google Gemini and is intended to be used as a target in an operation.

  • Prompt: Sends a prompt to the specified model in Google Gemini and is intended to be used as a target in an operation.