Users of BSL usually work with dbt.
We want to help them generate a scaffold of the semantic model (SM) by sourcing their existing dbt project.
The goal is to produce a v0 of the BSL YAML file.
We should use as sources:
• a dbt model selector to only target relevant models
• the .yml file of each model to extract: PK, column descriptions, and joins
• automatically add varchar columns as dimensions and numerical columns as measures
Questions:
• Should this be handled with a simple LLM prompt?
• What should happen if the underlying dbt models are updated?
Users of BSL usually work with dbt.
We want to help them generate a scaffold of the semantic model (SM) by sourcing their existing dbt project.
The goal is to produce a v0 of the BSL YAML file.
We should use as sources:
• a dbt model selector to only target relevant models
• the .yml file of each model to extract: PK, column descriptions, and joins
• automatically add varchar columns as dimensions and numerical columns as measures
Questions:
• Should this be handled with a simple LLM prompt?
• What should happen if the underlying dbt models are updated?