Databricks is working hard to make data more accessible to business users.
In the catalog explorer users can directly interact with the data using a LLM. Questions asked are interpreted into SQL and deliver insights.
This works quite well even for more complex tasks. I asked for the average number of days between shopping trips and it came up with a reasonable solution with multiple steps.
But as with anything in data, having metadata like table and column comments improves the results a lot because the LLM does not have to do the guesswork.
