Better column descriptions made the AI right more often. Not reliably.
I assumed most "the AI can't query our warehouse" problems are really documentation problems. Before building an evaluation harness to test that properly, I ran a small experiment.
Half right, it turn
zhaidata.com6 min read
Puneet Khandelwal
Turns out context actually matters more than raw parameter count. Garbage in, garbage out still applies to LLMs.