Glossaire
AI scoring
Cet article n'est actuellement disponible qu'en anglais.
Using a trained model to estimate the risk of an application from the data already gathered about it.
An AI score turns the material collected during onboarding — verified identity, bank transaction behaviour, screening results, application details — into a risk estimate and a recommendation. It does not replace the credit policy; it prioritises and informs the decision made under it.
A model trained on your own historical portfolio reflects your customers and your definition of a bad outcome. Without that history, the sensible path is to agree the criteria explicitly, start with those rules, and retrain on real outcomes as they accumulate.
Whatever the model says, the reason has to be recordable. A decision a lender cannot explain is a decision a regulator will not accept.
Sur la plateforme
Termes associés
Un système exactement comme vous l'imaginez
Parlons-en. Il est temps de donner une meilleure version à votre activité.