
Research question 01
Blind evaluation
How close can AI get to a buyer's reaction, and where does it break?
We are comparing simulated buyer reactions with blind human evaluations to learn where they agree, where frontier-model baselines fail and what remains irreducibly human.