The Airflow DAG is being "eaten" from the inside out, with specialized tools outsourcing computations and self-managing orchestration in their own domain. This has led to a constellation of tools, each with its own logical model of assets or entities they manage. Unbundling creates operational problems, including an operationally fragile data platform that leaves stakeholders confused about what ran and when. To address this, Dagster 0.14.0 introduces software-defined assets, which enable the authorship of these assets directly in Python and provide a unified view of all data assets in a single surface. This "rebundles" the platform by ingesting assets into a single control plane that provides lineage, observability, and data quality monitoring for the entire platform. Software-defined assets change the mental model of orchestration, enabling a declarative approach to asset management and providing a unified view of all data assets in a single graph.