Fukuoka Cafe Tests TASTE SIGNAL, an AI Platform for Sensory Data
The experiment targets a gap the company describes in food and fragrance development: sensory judgments like sweetness or roasted aroma stay qualitative unless someone measures and accumulates them, so generative AI has no grounded data to work from.
Reporting from 1 source: ASCII.jp.
Circlia Inc., which runs Holland Village Cafe FUKUOKA and the tea brand Tea by Circlia, has begun a demonstration experiment of TASTE SIGNAL. Customers taste products such as a cookie tin and rate them on five axes: sweetness, saltiness, texture, roasted aroma, and aftertaste, plus a four-point repeat-purchase question. The generative AI then proposes directions for the next prototype. The cookie tin sold out its 100 tins the same day.
TASTE SIGNAL collects evaluations in a common format so they can be compared and aggregated rather than left as scattered impressions. The accumulated data is split between groups with high and low repeat purchase intention, which makes differences in how each group rated a product visible. From there, the generative AI proposes directions for the formulation of the next prototype.
Circlia describes the current stage as verbalizing hypotheses from that accumulated data. The stated goal is a continuous cycle of tasting, sensory evaluation, purchase intention, formulation change, and re-evaluation, including how purchase intention shifts when a formulation is altered.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.