Databricks launches Governance Hub for account-wide tracking

Databricks has introduced Governance Hub in public beta, a centralized dashboard designed to give platform teams and administrators a unified, account-level view of their entire data estate. Operating across Amazon Web Services, Microsoft Azure, and Google Cloud Platform, the new hub aggregates information that was previously scattered across system tables and individual workspace views. It aims to streamline operations for data governance, financial operations, and AI deployment teams without requiring them to build custom dashboards.

The platform is divided into distinct verticals, starting with a Data page that highlights metadata gaps by showing the percentage of assets that are tagged, owned, and classified. Practitioners can drill down into specific schemas to find missing descriptions or identify unhealthy metastores. Additionally, an Access Insights feature provides a single, principal-centric view of permissions. This allows administrators to select any user, group, or service principal to instantly audit their direct and inherited access privileges across the entire account.

For organizations deploying machine learning, the AI page monitors traffic flowing through the Unity AI Gateway. It tracks token consumption, model activity, per-user spending, and guardrail coverage, alerting administrators when users exceed set thresholds. Meanwhile, the Cost page tracks financial metrics including 30-day spend, month-to-date totals, and daily averages. It specifically highlights untagged resources, helping financial operations leads identify untracked expenditures that escape standard budgeting processes.

To simplify querying, Databricks has integrated its Genie conversational agent into the hub. Users can ask natural language questions about cost spikes or security policies to receive recommendations without writing SQL. Currently, account administrators can enable the beta feature from the Account Console. The hub respects existing permissions, meaning workspace administrators only see cost and AI data for their workspaces, while metastore administrators are limited to data metrics.