Case Study

Virtual Digital Assistant

Belfius Bank

With more than 3.5 million customers, Belfius Bank is the second largest bank and insurance company in Belgium.

The Challenge

An increasing volume of customer queries led Belfius Bank to seek a digital solution to alleviate the rising demand on their call centres. In pursuit of round-the-clock accessibility, and to enable automated responses to “easy” questions, Belfius decided to start the development of ‘BBot’; a Virtual Digital Assistant, powered by Machine Learning (ML). In order to handle the vast array of questions and remarks from Belfius Bank’s diverse customer base in a more efficient, responsive and customer friendly way, BBot required two modules: a chatbot interface platform, and a Natural Language Understanding (NLU) engine, enabled by ML. Belfius Bank chose to partner with ML6 to build the NLU engine.

The Solution

The NLU engine that ML6 built for Belfius Bank is fully integrated within the BBot chatbot interface (which was developed by another of Belfius’s partners). Deployed across a number of platforms - including social media and the Belfius app - the NLU engine is able to understand customers’ requests and extract the appropriate information (such as bank card type) in Dutch and French. To enable this, ML6 developed a neural network for ‘intent classification’ which identifies the reason for the query, as well as ‘entity extraction’ which identifies the particulars of the query, such as location and date.


With an accuracy rate exceeding 90% in both Dutch and French, the NLU engine is successfully reducing the volume of customer queries being directed to Belfius Bank’s call centre.