Articles in this section

Set up a connection to Google Vertex AI

Naga Sri Hari Gonaboyina
Updated

Google Vertex AI is a unified machine learning platform that lets developers build, deploy, and scale AI models and generative AI applications on Google Cloud.

After you start creating an AI agent and select Bring your own key (BYOK), select an existing connection or create a new one. If you create a new connection, complete all of the required * authentication settings described below:

Setting Instructions
Name your connection *

Enter a clear and distinguishable name.

Throughout integrator.io imports and exports, you will have the option to choose this new connection. A unique identifier will prove helpful later when selecting it from a list of the connections in your account.
Service account key (JSON) * Paste the entire downloaded file into the Service account key (JSON) field, including the -----BEGIN PRIVATE KEY----- / -----END PRIVATE KEY----- lines. For information on downloading the key, see Keys.
Project ID *

Enter your Google project ID. You can identify your project ID in the JSON key as shown in the following example:

{
  "type": "service_account",
  "project_id": "my-company-ai-prod",
  "private_key_id": "0123456789abcdef0123456789abcdef01234567",
  "private_key": "-----BEGIN PRIVATE KEY-----\nMIIEvQIBADANBgkqhkiG9w0BAQEFAASCBKcwggSjAgEAAoIBAQC...\n-----END PRIVATE KEY-----\n",
  "client_email": "vertex-agents@my-company-ai-prod.iam.gserviceaccount.com",
  "client_id": "123456785678901",
  "auth_uri": "https://accounts.google.com/o/oauth2/auth",
  "token_uri": "https://oauth2.googleapis.com/token",
  "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
  "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/vertex-agents%40my-company-ai-prod.iam.gserviceaccount.com",
  "universe_domain": "googleapis.com"
}
Location * Enter your Google Vertex AI account region. For example us-central1, europe-west4, global. Make sure the model selected in the agent is available in the region. For information on the location, see Locations.

Tip

This connector documentation describes only the settings shown for the Simple view. For the corresponding HTTP settings, see the Custom auth universal connector documentation, and in Base URI, enter https://aiplatform.googleapis.com.

Points to remember

  • The Project ID field must match the project_id value in the key file (my-company-ai-prod in this example).
  • Only service account keys ("type": "service_account") are accepted. User-credential files (authorized_user) and workload-identity files (external_account) are rejected.
  • The service account must have the Vertex AI User role (roles/aiplatform.user) on the project.
  • The key is stored encrypted and cannot be viewed again. If the connection is edited later, the key must be pasted in again.

Additional references