Google Cloud has launched Knowledge Catalog, a platform designed to provide context for enterprise AI agents. The service, which debuted on April 22, 2026, is a rebranded and upgraded version of the former Dataplex Universal Catalog.
The system functions as a context engine that integrates technical and semantic metadata from Google Cloud services and partner platforms. It utilizes AI-driven insights to provide governed context, aiming to reduce hallucinations and improve the accuracy of AI agents.
The platform supports integrations with Snowflake, SAP, and Looker, allowing for unified semantic context without the need for data replication. These integrations utilize zero-copy access, enabling the catalog to read data meaning while the information remains in its original location.
In October 2026, Google introduced self-learning context enrichment and an updated Knowledge Studio, both currently in preview. Additionally, LookML semantic language indexing for Looker Core reached general availability, enabling live synchronization of business logic.
Google reports internal benchmark results for agents using the Knowledge Catalog, claiming nearly four times greater accuracy, up to 33% lower latency, and a 75% decrease in token usage.
