When you guys are developing some model in snowflake or any other warehouse, do you start by getting your query logic fully figured out in your data warehouse interface (like ssms/dbeaver/whatever) and then just copy over that code to your dbt model? or do you do everything through the dbt command line?

Sometimes I start in the db client and then copy that to the dbt model (if it’s a brand new model with no dependencies). But my usual workflow is to start with the dbt model definition in my IDE and then run

dbt compile

and run that in the db client as I work on the model.