Digital Café - Zoe.ai - Behavioral Prescoring

02.08.2019

At the July Digital Café we introduced our product Zoe.ai - Behavioral Prescoring, which helps with client risk assessment in the online loan application process. And it can do much more.

Lundegaard Digital Café is a platform for meeting experts from Lundegaard and the related industries individuals. The aim of the meeting is to discuss new approaches and experiences with building real digital services projects. This time the event brought together people from the insurance and banking sectors.

Jan Procházka (Director of Digital Solutions & Consultancy, Lundegaard) gave the opening speech, outlining the morning agenda, and explaining why Lundegaard embarked on the path of fast data. “We have been building front-ends for our clients for over twenty years and have seen what digital footprint people leave behind and how much data is generated in this environment. So we naturally got into the development of services in the area of fast data and artificial intelligence, ” Jan said.

In his presentation, Jan Denemark (Senior Consultant, Lundegaard) focused on the main problems with the scope of the use of artificial intelligence and fast data. “The three main reasons why AI and fast data are not being used much yet are the following: First, we don't know where to use AI and fast data to make it useful for business. Secondly, we are not easy to define the main use cases in a given area. And finally, we struggle to choose the relevant technology from the plethora of technologies.” Jan tried to address these topics, which also sparked a very interesting discussion with the participants.

Lukáš Matějka (Head of Fast Data and Artificial Intelligence, Lundegaard) talked about our product Zoe.ai - Behavioral Prescoring. First, he introduced our Brno Fast AI team, which has been developing this competence, i.e. a micro-service platform that uses fast data and artificial intelligence to develop business data-intensive applications. The platform is built on Kafka, Kafka Streams, and H2O.ai open-source technologies.

Zoe.ai is based on this very platform, monitoring detailed online user behavior in the online environment during the process of getting a loan (typically through some type of an online application form), and following this provides the user with summary information and predictions, which are based on algorithms and machine learning.

The main advantages of this product include secure data acquisition, which does not increase the demands on the loan applicant. And thanks to better client risk assessment, the loan can be approved or rejected faster, saving time and resources. Based on evaluating the data and finding the profiles of ‘good’ borrowers, more efficient targeting of online offers and ads to potential site visitors can be made.

Lukáš closed his talk with practical demonstrations, where he showed how the product works in practice. “As one of the many indicators about the user, we also use the so-called ‘key stroke dynamics’ method, i.e. mapping the applicant’s hesitation, rewriting of fields, how quickly she writes or copies the text in each field. For example, if she rewrites her name several times, we can take it as a sign of attempting to make a fraudulent application,” Lukáš explained as one example of what Zoe.ai can do.

Lukáš’s presentation was followed by our guest from Cofidis, Milan Bouda (Head of Data Analytics), who talked about using Zoe.ai in practice. Among other things, he mentioned that Cofidis's goal is to request from customers as little data as possible, while making the best loan decisions, which also made them opt for behavioral prescoring technology. Zoe.ai helps them predict loan approval, early repayment or fraudulent behavior.

Another guest and the last speaker was Petr Zapletal (ex-CEO and Chairman of the Board, Pojišťovna České spořitelny), who explained why prescoring is also useful in the insurance world and introduced a modification of Zoe.ai for the insurance industry. The three main benefits of the product include risk assessment - finding a risky client, increasing sales - the ability to create better deals for certain clients and ultimately fraud prevention.

More about Zoe.ai at zoe.lundegaard.ai.

Thank you all for attending the event, for the interest and fruitful discussion that took place after the presentations. We already look forward to meeting you again to have a good breakfast at the next Digital Café.

Digital Café - Zoe.ai - Behavioral Prescoring
Digital Café - Zoe.ai - Behavioral Prescoring
Digital Café - Zoe.ai - Behavioral Prescoring
Digital Café - Zoe.ai - Behavioral Prescoring
Digital Café - Zoe.ai - Behavioral Prescoring