genie-space-patterns

To configure Databricks Genie Spaces for production, implement a mandatory 7-section deliverable structure, optimize agent instructions to twenty lines or fewer, and organize data assets hierarchically using Metric Views, TVFs, and Tables.

The best way to structure data assets in a Databricks Genie Space is to follow a strict hierarchy: prioritize Metric Views, followed by Table Valued Functions (TVFs), and finally base Tables, ensuring each asset meets documentation standards.

Validate benchmark questions for Databricks Genie Spaces by executing an interactive intake workflow that tests natural language queries against the configured agent instructions and data assets to ensure accurate analytics responses.

Databricks Genie Spaces may return inaccurate results if General Instructions exceed twenty lines or lack conciseness, or if data asset organization fails to follow the Metric Views, TVFs, and Tables hierarchy defined by best practices.

Using natural language analytics in Databricks without organizing data assets hierarchically is not recommended, as production-ready Genie Spaces require strict adherence to Metric Views, TVFs, and Tables configurations for accurate query responses.