answering-natural-language-questions-with-dbt

You can answer business questions with dbt by querying the Semantic Layer first to retrieve defined metrics, falling back to ad-hoc SQL to handle gaps and return executable queries for your data warehouse.

The best way to query dbt metrics for ad-hoc data analysis is to check the Semantic Layer first, then use ad-hoc SQL as a fallback to surface actionable insights without building new dbt models.

You do not need a fully defined Semantic Layer to analyze data with dbt. The system applies a Semantic Layer first approach, but falls back to ad-hoc SQL to handle gaps in the semantic model.

When metrics are missing from the model, dbt handles business questions by falling back to ad-hoc SQL, using manifest analysis and model discovery to surface relevant artifacts and data sources for the query.

You can use dbt to discover data sources for ad-hoc SQL queries by performing manifest analysis, which surfaces relevant dbt artifacts and data sources to provide scalable and reusable responses.