I'm using LLMs to help build dbt-flavored Great Expectations tests. Pretty low-hanging fruit, but it's a start.