Regionalization and data residency

Agent Assist provides data residency to keep your data-at-rest physically within a specified geographical region or location. When you specify a region, your data-at-rest is not stored outside that region. Pricing is the same for all regions.

Specifying a region is particularly helpful in the following use cases:

  • Your system has regulatory or policy requirements that govern where your data must reside.
  • Your network latencies might be improved when the data is in the same region as your customers. For example, if Belgium customers use europe-west1, they can expect better latency.

Available regions

Country grouping Geographical location Region ID Data store region ID
Americas Iowa us-central1 us (multi-region)
Americas South Carolina us-east1
Americas Oregon us-west1
Americas Montréal northamerica-northeast1
Americas United States multi-region us * us (multi-region) *
Europe Belgium europe-west1 eu (multi-region)
Europe London europe-west2
Europe Frankfurt europe-west3
Asia Pacific Singapore asia-southeast1
Asia Pacific Tokyo asia-northeast1
Asia Pacific Mumbai asia-south1
Asia Pacific Sydney australia-southeast1
Global Global multi-region (global serving, data-at-rest in United States) global (preferred) or no region (default) ¶ global (multi-region) ¶
Symbol Description
* Vertex AI does not support the 'us' multi-region location. As a result, using Agent Assist's Generative AI features in 'us' multi-region will rely on the respective existing US single region endpoints.
Calls to the global region might have higher latency when called from outside of the US. Prefer a more specific region that is closer to your services and users, if possible.

Available features for regional serving

Region ID Summarization v2 custom model (preview) Summarization v1 baseline model Summarization with custom sections Sentiment Analysis (Proactive) Generative Knowledge Assist
us-central1 ✓ ✓ ✓ ✓ ✓
us-east1 ✓ ✓ ✓ ✓ ✓
us-west1 ✓ ✓ ✓ ✓ ✓
us ✓ ✓
northamerica-northeast1 ✓ ✓ ✓ ✓
europe-west1 ✓ ✓ ✓ ✓
europe-west2 ✓ ✓ ✓ ✓
europe-west3 ✓ ✓ ✓ ✓
asia-southeast1 ✓ ✓ ✓ ✓
asia-northeast1 ✓ ✓ ✓ ✓
asia-south1 ✓ ✓ ✓ ✓
australia-southeast1 ✓ ✓ ✓ ✓
global ✓ ✓ ✓ ✓ ✓

Data-at-rest

All Agent Assist developer user and end-user data is included in data-at-rest. For example:

Limitations

The following limitations apply:

  • The Agent Assist Console does not yet support regionalization. You can only implement regionalization by calling the API directly.
  • Model training does not support regionalization. Your data may get routed outside the region during this process.

  • API Explorer, found on many REST reference documents, only supports the global region for API calls.

  • CCAI Transcription currently only supports multi-region data in use and data at rest (if not using the Speech Adaptation feature) in EU, US, North America (Canada).

  • Only the following regions are supported for data stores, used by (Proactive) Generative Knowledge Assist feature: global, us multi-region, and eu multi-region.

Specify a region using the API

To use regionalization you must specify a region for every API request.

Required fields:

  • Provide the region ID for the location URL path parameter.

  • For the global region, use dialogflow.googleapis.com for the hostname. For all other regions, use [region-id]-dialogflow.googleapis.com for the hostname.

For example:

  • dialogflow.googleapis.com/v2/projects/project_id/locations/global/knowledgeBases/knowledge_base_id

  • us-central1-dialogflow.googleapis.com/v2/projects/project_id/locations/us-central1/knowledgeBases/knowledge_base_id

To use client libraries, the following steps are required:

  • Set the service endpoint to: REGION_ID-dialogflow.googleapis.com.

  • Set the resource name with the correct location, for example:

    projects/project_id/locations/region_id/knowledgeBases/knowledge_base_id

Credentials

We recommend that you use the GOOGLE_APPLICATION_CREDENTIALS environment variable to provide your client libraries with credentials. See the Dialogflow setup instructions for more information about setting this variable.

If you instead need to provide manual credentials in your code, one additional step is required before you can call a regionalized endpoint. You must set the scope to https://www.googleapis.com/auth/cloud-platform when you generate your authentication credentials. The following example demonstrates how to do this in Java when creating a GoogleCredentials object:

    GoogleCredentials.fromStream(new FileInputStream("credential_file"))
      .createScoped("https://www.googleapis.com/auth/cloud-platform")

Cloud Logging

See the Cloud Logging documentation to control the region where your logs are stored.