Hi @john77 , 

When you have a SQL task creating ST/MV ,it works fine for a few independent tables. You do get incremental refresh, retries, and an auto-created (implicit) pipeline per object
But what you miss is that ,it is  Harder to mix SQL + Python or add things like AUTO CDC and expectations across the whole pipeline.

No single DAG that guarantees bronze → silver → gold(medallion) ordering as one unit and object has its own run/metrics; no one “pipeline run” view.

All the above are only possible with ETL pipline