Who asks, who records: separate model roles in a first-conversation assistant
Models ask in similar ways but record differently. Briefs written separately by O-05 and O-06 kept 27 of 27 key facts, according to the rating models.
ReadI want to know why it works, under what conditions and when it fails.
Alongside client projects, I run my own experiments with systems, data and technology. Here I publish selected observations, results and conclusions — including when they overturn my earlier assumptions.
Bartłomiej Pietrzyk
4digital Lab
No publications match your search.