Agentic AI Finance: Governed Metrics, Open Access, Automated Action on Databrick

Large finance organizations already rely on dashboards and reports, but the real challenge is enabling automation without breaking governance or auditability. This session shares how a global finance organization implemented agent-driven workflows on the Databricks Data Intelligence Platform using Unity Catalog, metrics views, and open lakehouse patterns.

The talk walks through a real implementation where finance metrics are defined once using metrics views, protected with attribute-based access control and automated classification, and reused consistently across reporting, analytics, and automation. Power BI connects directly to governed lakehouse tables and metrics for enterprise reporting, while Iceberg-compliant clients consume Delta tables through Uniform, enabling open access without duplicating data.

Agentic workflows observe certified metrics and events, assist with variance investigation, anomaly triage, and operational follow-ups, and propose actions that remain traceable and reviewable. Early results indicated several hours of manual investigation effort saved per reporting cycle and faster anomaly triage, without introducing additional governance or audit risk. The session includes an architecture walkthrough and demo using synthetic data, reflecting real finance constraints, trade-offs, and lessons learned when moving from reporting to governed, open, and action-oriented analytics.