With its diverse range of insurance and financial services, the Debeka Group is one of the top five in the German insurance and home savings industry. Every year, Debeka receives around 16 million document pages via a wide variety of input channels: classically by mail or fax, by e-mail, as an upload via the customer portal, or via sales. In addition to the sheer volume of documents, the complexity of incoming document types is a challenge due to the many different insurance lines.

Debeka has been using our smart FIX solution very successfully for several years. Essentially, two use cases were implemented here: identifying the responsible division and analyzing the document structure and content. Based on the identified document class, routing to the appropriate service department takes place and the corresponding workflows are started.

The incoming mail of the car insurance and building society divisions is characterized by many incoming documents with images such as accident pictures, real estate plans, or ID documents. These documents offer little potential for purely text-based classification and thus had to be indexed almost entirely manually. Thus, the idea arose to extend the text classifier used so far with an image classification. The new Cognitive Classifier now combines the best of both worlds: the proven excellent text recognition and now, newly, the classification of images with Deep Learning. The previous workflows remained unchanged and the front end for users also remained the same. The AI continues to work reliably in the background and decides whether text recognition, image recognition, or both are to be used.

The automation effect of the combined classifier is enormous. While the majority of documents containing images had to be indexed manually before the introduction of the new process, this is now less than 10 percent.

“Together with our partner Insiders Technologies, we have deployed an AI solution that is absolutely at the cutting edge for an important use case.This makes working with Insiders fun!”

says Patrick Schneider, Head of Frontend Department at Debeka Krankenversicherungsverein a. G.

With the Cognitive Classifier, a universal solution for the classification of documents and photos is now available that puts all existing approaches in the shade. This allows proven processes to be used even more effectively and the level of automation to be further increased. The insurer’s and Insiders’ project for the Cognitive Classifier is already attracting a great deal of interest in the industry.