Writing robust data pipelines with AI

​AI writes data pipelines in seconds.

​The catch? A pipeline that runs green and a pipeline that returns the right numbers are two very different things. AI is non-deterministic and can't see your data, so it guesses your schema, your units, your duplicates, and those guesses become silent bugs no error will ever flag.

​In this session, Mehdi shares the practical habits that turn a lucky one-shot prompt into a pipeline you can actually rely on with live examples using Claude Code and MotherDuck, but every pattern maps to your own stack!