My team recently finished architecting an Agentic GenBI pipeline (we’re using dbt to build a governed semantic layer that feeds into a Python/LangChain reasoning loop). The AI side is working beautifully, but we hit a frustrating snag with our Azure SQL materializations.
When executing a standard dbt run, it frequently fails to cleanly overwrite existing views in our Azure SQL environment.
Our current “hack”: We are manually dropping the view in SQL before running dbt to force a clean build.
I know this is a massive anti-pattern. Has anyone found the correct adapter config, macro, or dbt_project.yml setup to make dbt run handle view overwrites gracefully in Azure? Any pointers would save us a lot of headaches!