Dufry | CADMUS
case dufry

Duda, the robotic saleswoman with a
back office made for conversational services

Travelers who pass by one of the Dutty Free Brazil stores, in eight of the country's main airports, can consult, reserve products or close the purchase, with the help of Duda, the virtual assistant that attends through the chat channels.

Architecture focused on Artificial Intelligence

The innovation combines Neomode's ShopBot and a framework developed by Cadmus, which optimizes the storage, search, and delivery of information for Duda to do her job. To adapt the data structure and functionalities to the dynamism of the conversational interface, Cadmus built the solution with MongoDB and Node.js.

With a document-oriented architecture, similar to that of search portals, the data structure that serves the chatbot allows a more natural modeling of the information for the virtual assistant.

Continuous Improvement

Since this is a project in constant improvement process, scalability and the ease of designing new query models is another premise. In this context, the Node.js platform was adopted - typically used in companies focused on the search experience and contextualization of data, such as Netflix - a Java Script execution environment on servers.

Together with the work with Dufry's support team, the natural interaction intelligence of the Neomode and IBM Watson platforms, Cadmus guarantees that Duda has more and more assertiveness so that no consumer fails to find and buy what is interesting to them.


Let's go on this journey together?

We turn your ideas into reality, achieving results
with the use of agile methodologies in your business.