dbt shouldn't be running in a silo. It's important that your models are processed immediately after the data they relate to is processed. With Shipyard, you can an interconnected workflow that uploads data to a table in PostgreSQL and immediately triggers a subset of your dbt models. This way, you can guarantee that your dbt models are only running once new data is loaded. If the PostgreSQL data runs into trouble, you can prevent the downstream dbt job from ever running.