Gateco Now Supports Google Vertex AI: Extending Policy-Aware Retrieval to GCP

We're excited to announce that Gateco now integrates with Google Vertex AI Vector Search, making it our 11th supported vector database connector and our first native GCP integration.

Vertex AI Vector Search is Google's managed approximate nearest-neighbor service — a high-performance, low-latency vector retrieval engine designed for large-scale embedding workloads. Teams using Google Cloud for their AI infrastructure can now apply Gateco's full policy enforcement stack to their Vertex AI indexes without re-indexing or modifying their data pipelines.

What This Means for Your RAG Pipelines

Until now, connecting a Gateco policy to a Vertex AI Vector Search index required routing retrievals through an intermediary or maintaining a separate Gateco-managed copy of your vectors. With the native connector, Gateco queries your existing Vertex AI index endpoints directly — then applies RBAC and ABAC policies to filter results before they reach your AI agent or application.

Every retrieval is logged to Gateco's audit trail with the full policy decision context: which principal made the request, which policies were evaluated, which results were allowed or denied, and why. Nothing bypasses the policy layer.

How the Integration Works

Setup requires four pieces of configuration: your GCP project ID, the index endpoint ID, the deployed index ID, and your index's public endpoint domain. Authentication uses a GCP service account with the Vertex AI User role — the same service account pattern you're likely already using for other Vertex services.

Because Vertex AI Vector Search is a pure vector store (it doesn't index raw text alongside embeddings), Gateco uses sidecar metadata resolution by default. Your existing metadata — document titles, classification labels, owner groups — is read from Gateco's own metadata store and attached to each retrieval result before policies are evaluated. No changes to your Vertex index are required.

Tier 2 with Full Policy Coverage

The Vertex AI Vector Search connector is classified as Tier 2: it supports vector search and policy enforcement, but Gateco does not manage ingestion — your existing pipelines continue to write vectors to Vertex AI directly. Gateco attaches metadata and policies to the resources retroactive registration isn't needed; resources are registered via Gateco's API or dashboard as part of onboarding.

Despite being Tier 2, the connector supports the full Gateco policy stack: RBAC rules, ABAC conditions on resource metadata, deny-by-default enforcement, the Access Simulator for dry-run testing, and real-time audit logging for every retrieval decision.

Vertex AI Search Is Next

We're currently building support for Vertex AI Search — Google's fully managed enterprise search service with native keyword, vector, and hybrid retrieval. Vertex AI Search adds ranked keyword search and native RRF hybrid fusion on top of the vector capabilities, making it comparable to Azure AI Search in the Gateco connector tier hierarchy.

Vertex AI Search is now available as connector type `vertex_ai_search` and supports retroactive registration via its Documents API. Connect it from the Gateco dashboard today.

Getting Started

Vertex AI Vector Search is available in Gateco today on all plans (Free, Pro, Enterprise). Connect your first Vertex AI index endpoint from the Connectors page in the Gateco dashboard, or use the Python or TypeScript SDK: `client.connectors.create(type="vertex_ai_vector_search", config={...})`.