No way to hard-stop Agent Search (Vertex AI Search) on budget/quota threshold?

Google Cloud Build with AI Agents
1

We are deploying Agent Search (formerly Vertex AI Search) on a public
corporate website using the embedded gen-search-widget. Billing is
usage-based, and our client requires an automatic stop mechanism in case
of unexpected traffic spikes such as bot attacks.

What we have already verified:

  1. Cloud Billing “spend cap enforcement” (Preview) does not support this
    service. The service picker in that mode lists only 56 services, and
    Vertex AI Search (service ID 74B1-77CF-C302) is not among them.
    In “alert only” mode, 1,777 services are listed and Vertex AI Search
    IS selectable - but alerts only notify, they don’t stop anything.

  2. Setting the “Search lite requests per minute” quota to 0 does not stop
    the widget. The widget appears to consume “Regional search requests
    per minute per regional_location”, which is a system limit and cannot
    be adjusted (300/min for global).

  3. Our SKU shows: Vertex AI Search: Search API Request Count - Enterprise,
    free up to 10,000 queries/month.

Questions:

  • Is there any supported way to hard-stop Agent Search when a monthly
    query threshold is reached?
  • Is disabling the Discovery Engine API via Cloud Function (triggered by
    a Monitoring alert) the recommended pattern for this?
  • Are there plans to add Vertex AI Search to spend cap enforcement?

Any guidance appreciated.

1 Like