Leave orchestration to the engineers, and let the business feast on its metadata
Great post @louise! A natural question for me is -- what do you think about the reverse formulation -- that the orchestrator could eat "active metadata" ? Dagster, dbt Cloud, and Airflow, to name a few examples, all collect a lot of "asset / model / dataset" metadata and make it available via APIs. If they began exposing business-user-friendly interfaces, do you think they would be competitive with active metadata tools like Castor and Atlan?