Most teams have text columns full of answers they can't GROUP BY: agent traces, comments, job ads, call notes.
In this session, Mehdi Ouazza walks through five real use cases built with MotherDuck's prompt_jev(), from AI agent logs to Hacker News social listening and job ads. Each one answers a business question in a single SQL query, and Jev runs about 20 times faster than a standard LLM call, at a fraction of the cost.
You'll leave with the SQL, public shares you can query yourself, and a checklist for your first experiment.
