In a 25-request shopping study, Instinct led on writing and Dots on product fit

Shopping agents choose products as well as explain them. Braintrust, an AI evaluation company, tested whether those two jobs were being done equally well.

What happenedIn Braintrust's 25-request live-web study, one blinded reviewer ranked Instinct first for written answers and Dots first for product fit. The chart below separates those two judgments.

Why it mattersA persuasive explanation can hide a weak choice. The automated live-web score measured what the saved answer documented, not a fresh check of every retailer page; its ranking tells a different story again.