Fraud detection


This module is used to detect and prioritize suspected fraud cases by taking into account all the information on the settlement files. It adapts perfectly to databases that do not have a lot of verified fraud data. The solution is anything but a "black box", it allows you to have explanations for each suspected case of fraud detected. Intuitive visual reports showing cases of fraud over time and geography are also customizable.

Detection, Extraction and Classification of Information in Documents

This module is able to recognize and extract information from different types of documents: Passports, Identity Cards, Contracts, Orders, Purchase Orders or Invoices. The platform encapsulates deep learning algorithms capable of recognizing and extracting information such as First Name-Last Name, Addresses, Cost Centers, and Signatures. When the information is extracted, it can be linked to an existing database to enrich or validate certain fields.

Data reconciliation

Companies with a significant history have a multitude of different information systems with dispersed and often seperated tools and databases. This module makes it possible to reconcile records of different databases thanks to several fields without necessarily having a common identifier. The solution uses multi-field matching based on numeric and textual variables using syntaxic and semantic analysis. The optimal match between records evolves over time, thus taking into account the quality of the data. You can, through a very simple interface, choose the reconciliation axis for immediate or recurrent reconciliation between one or more databases.