Deployment and analyst-facing tooling around the Databricks platform.
The tools in this section sit at the two ends of a Databricks project. At the engineering end, Databricks Asset Bundles package jobs, pipelines, notebooks, and their configuration as code so they can be reviewed, tested, and deployed through CI/CD to separate environments. At the consumer end, Genie and AI/BI dashboards let analysts and business users ask questions of governed Unity Catalog tables in natural language or through dashboards, without writing Spark code.
The two tools meet at Unity Catalog. Bundles deploy the jobs that produce well-defined tables; Genie and dashboards read those tables under the same permissions, so the quality of the curated layer and its column descriptions directly determines how good Genie's generated SQL is.
Asset Bundles are driven by the Databricks CLI: databricks bundle validate checks the manifest, databricks bundle deploy -t prod pushes resources to a target workspace, and databricks bundle run triggers a job. Targets override variables such as catalog names and cluster sizes, which keeps one definition across environments. Genie converts a question into SQL against a curated set of tables, runs it on a SQL warehouse, and shows the query so results can be checked; instructions and example queries steer it toward correct joins and business definitions. AI/BI dashboards provide the fixed, shareable views for metrics people check repeatedly.
| Need | Tool |
|---|---|
| Version-control and promote jobs across environments | Asset Bundles |
| Ad hoc questions from non-SQL users | Genie |
| Recurring KPIs shared with a team | AI/BI dashboards |