Verdify and Genzai will integrate a range of innovations in the new Verdify 2.0 platform that will be launched next spring. Among these innovations is taste prediction. Based on extensive research, discussions with multiple taste scientists and interdisciplinary brainstorms, we have developed a model that can be used to describe the taste of any recipe. Thus far it is applied on the 1 million recipes in our database, but soon it will be applicable to any recipe on the internet.
Such a model is not only useful for communicating the taste profile of a meal to consumers, but also for creating a Netflix- or Spotify-like experience. By collecting and analyzing user behavior on a recipe platform – such as number of views, ratings and conversion to shopping lists – our systems can automatically assemble an individual taste silhouette of consumers. This translates in highly personalized recipe suggestions whenever a consumer searches for meal inspiration. Probably liked recipes can be shown more frequently than those with a less close match between recipe profile with taste silhouette. Any Verdify user is thereby offered a unique digital experience that is fully adjusted to his or her nutrition profile ánd taste preferences. The prediction accuracy will improve over time due to the self-learning nature of the algorithms.
This is the fifth snapshot by Verdify and Genzai in a series that proceeds over the next months. Each week we publish a sneak peek into the A.I. recipe solutions that we co-develop.