A feature (not a skill) inside Higgsfield's Supercomputer that scores finished content 0-100 by modeling brain responses across vision, sound, and memory, then breaks the score down into hook analysis, pacing analysis, and estimated retention-curve analysis, with actionable fixes attached to each component.
The core idea: instead of publishing and waiting roughly a week to learn whether a piece of content performs, run it through a predictive scoring model first and get an instant, actionable read plus specific fixes. It generalizes past Higgsfield's own ad/UGC use cases — the source notes it applies "no matter if it's cinematic shots or just a cartoon" — and works either as a standalone pre-publish check or as the numeric pass condition driving Goal Mode: Closed-Loop Generate-Test-Select Content Production.