What Is dbt?

dbt (data build tool) is an open-source command-line tool that lets you transform data inside your warehouse by writing SQL SELECT statements instead of hand-rolled ETL scripts. Rather than extracting or loading data, dbt takes data that is already sitting in a warehouse like Snowflake, BigQuery, or Redshift and turns raw tables into clean, tested, documented models purely through SQL, compiled and executed as CREATE TABLE or CREATE VIEW statements against that warehouse.

Before dbt, the transformation logic that converts raw source tables into business-ready models often lived in a tangle of stored procedures, scheduled Python scripts, or opaque BI tool calculations, with no version control, no tests, and no shared documentation. dbt applies standard software engineering practices — version control with git, code review, automated testing, and modular reuse — to the analytics engineering workflow, so a SQL transformation is written once as a model, referenced by other models, and validated automatically before it reaches a dashboard.

dbt ships in two flavors: dbt Core is the free, open-source command-line tool you install with pip and run yourself, while dbt Cloud is a managed service built by dbt Labs that adds a browser-based IDE, a job scheduler, hosted documentation, and native git integration on top of the same underlying engine. Both compile the same Jinja-templated SQL into warehouse-native SQL and execute the same DAG of models; the choice is mainly about who manages the orchestration and infrastructure, not a difference in the transformation logic itself.