ReleaseBytes
gcp Google Cloud Blog ·

Looker's Semantic Layer Integrates with Gemini Enterprise for Governed AI Analytics

blogaigcpgaarchitectgcp-bigquerygcp-spannergcp-looker
feature announcement

Looker's governed semantic layer now integrates with Gemini Enterprise, providing a trusted foundation for structured data queries within AI agents. This integration aims to minimize AI hallucinations and ensure consistent, trustworthy analytics by routing requests through Looker's codified business logic. It enables organizations to publish conversational agents, access real-time analytics with rich visual interactivity, and maintain robust data governance via existing Looker permissions. The zero-risk pass-through architecture ensures secure access management without data ingestion or replication.

  • →Governed Semantic Layer for Gemini Enterprise
  • →Robust Governance and Secure Access Management
  • →Native, Interactive Charts in Gemini Enterprise
  • →Multi-Agent Interoperability via A2A Protocol
  • →Looker-Based User Authentication
Features (5) ›
  • Governed Semantic Layer for Gemini Enterprise

    The integration combines Looker’s semantic layer with Gemini Enterprise, allowing users to query both structured databases and unstructured documents in plain English from a single interface. This enables teams to instantly connect hard metrics with real-world context for faster problem-solving and decision-making.

  • Robust Governance and Secure Access Management

    The integration is built on a zero-risk pass-through architecture, processing data without writing to persistent storage and not ingesting or replicating underlying database records. It operates securely over the Agent-to-Agent (A2A) protocol, maintaining Looker’s existing row-level and column-level access controls and strict security isolation.

  • Native, Interactive Charts in Gemini Enterprise

    Looker agents within Gemini Enterprise now support rich visual interactivity, providing native, interactive data charts directly inside the universal chat box. When users request visual trends like monthly sales performance, the Looker agent maps the database response with presentation-ready visualizations. (Note: Looker agents published prior to release 26.12 may need updating for these capabilities.)

  • Multi-Agent Interoperability via A2A Protocol

    Looker agents published to Gemini Enterprise can understand context across different agents and data sources, leveraging standard communication frameworks to securely share governed insights. This enables complex multi-agent orchestration, where insights from Looker can feed into other first-party Google Cloud agents or external third-party agents for structured workflows.

  • Looker-Based User Authentication

    To preserve enterprise governance, the integration implements an identity-centric authentication model requiring a one-time OAuth consent from users. This binds their active Gemini Enterprise session to their Looker credentials, ensuring every conversational query is authenticated at the user level and enforces pre-existing Looker permission structures.

Enhancements (1) ›
  • Reduce AI Hallucinations with Codified Data

    Looker’s semantic layer eliminates guesswork for AI chatbots by providing critical context in the form of codified data, allowing Gemini Enterprise agents to generate deterministic and predictable responses. When a Gemini Enterprise user requests a KPI, the Looker agent generates precise SQL based on version-controlled business logic, ensuring consistent, governed metrics.

Read the original announcement →

https://cloud.google.com/blog/products/business-intelligence/integrating-looker-and-gemini-enterprise/

Related releases