Why a New Class of AI "Judgement Models" Could Have Big Business Implications

TypeSafe's new model Jev skips text generation entirely and returns calibrated probabilities instead, claiming speeds 20 to 200 times faster at a fraction of the cost. NLW explains what judgment models actually do, why so much business work is really just small decisions dressed up as text tasks, and where this fits alongside LLMs in a complete model stack. In the headlines: Zuckerberg pushing back on collective pacing, Bernie Sanders and Steve Bannon sharing a stage, and Salesforce shipping its first in-house model in years.