Yes, it’s viable — Databricks’ in-house agents can handle that workflow if you define each stage as callable Python tools. The key is robust function design and proper metadata access. However, you might find integrating Unity Catalog and Git integration APIs simplifies repo updates and version control.