Skip to main content
Connect to Gemini models served from your GCP project via Vertex AI, authenticated with a service-account key — the same mechanics as the BigQuery data connection. Steps to complete:
  1. Create a service account with Vertex AI access
  2. Configure in Datafold

​
Create a service account

In the Google Cloud Console, create a service account in the project that hosts your Vertex AI endpoints and grant it the Vertex AI User role (roles/aiplatform.user). Generate a JSON key file for it.

​
Configure in Datafold

  1. Navigate to Settings → Integrations, click Add New Integration, and choose Google Vertex AI.
  2. Upload the service account JSON key file.
  3. Models — the Gemini family is listed automatically from Vertex’s Model Garden. Partner models hosted on Vertex (e.g. Claude) use publisher-specific ids and can be added manually.
  4. Region (optional) — leave blank to use Vertex’s global endpoint, which serves mainstream Gemini models. Set a specific location (e.g. europe-west4) only for data-residency requirements or region-only models.
  5. Click Save.