In this chapter, you'll set up a Genie space and make it answer reliably by adding column descriptions, synonyms, default instructions, and entity matching.

Launching the Sales Pilot

Diagnosing Genie's default behavior

Adding descriptions and synonyms as insurance

Stacking default rules with instructions

Matching real values with entity matching

Next, you'll expand the space across multiple tables with relationships, and lock shared formulas using SQL examples and a parameterized Trusted Asset.

Department-Wide Adoption

Joining bookings for revenue by outcome

Segmenting customers by adding users

Locking the net revenue formula

Building a parameterized Trusted Asset

In this final chapter, learn to harden a Genie space using monitoring, targeted fixes, and a benchmark suite so its answers stay reliable at scale.

Enterprise Reliability

Diagnosing failures in monitoring

Locking the monthly trend formula

Defining "premium" with an instruction

Building a benchmark regression suite

Congratulations!

Databricks Genie answers data questions in natural language, but the quality of those answers depends on how the space is configured. In this case study, you take a Genie space through its full lifecycle. As a data analyst at Wanderbricks, you start from a minimal setup and build it out: defining column descriptions and synonyms, adding instructions, connecting tables, and saving example and parameterized queries. You then move into operations, using monitoring to surface weak answers and benchmark questions to measure quality. By the end, you can take a Genie space from setup to production-ready.

Introduction to Databricks Genie

Build a Databricks Genie space end-to-end: descriptions, synonyms, instructions, table relationships, example queries, monitoring, and benchmarks.

Defining "premium" with an instruction

Interaktive praktische Übung