Incident or Request?

Traditional BI Architecture

Every interaction between a customer and a supplier is either an incident or a request, and Alice figures it out automatically! Alice then figures out what service desk should handle the given ticket.

It’s Polyglot!

Alice reads, classifies, and routes any text in any language. Well, almost. Alice supports no less than 22 different languages.

It’s Service-minded

Traditional BI Architecture

Alice runs using web services. It means that is can be integrated with ServiceNow without impacting the performance or memory requirements of your ServiceNow implementation.

It needs training!

Traditional BI Architecture

Alice needs a set of messages or tickets as well as information on how they were categorised or routed. Once trained, it suits your specific purposes. With 4,000 entries, it is about 70% correct. With 8,000 entries, it gets about 80% right.

How does it work?

Traditional BI Architecture

Alice takes all of your previous tickets, reads their descriptions and creates a Machine Learning model. But it’s clever! It reads the sentences, delineates them and then feeds the model with the most important information. It make Alice better than the others.

Under the hood

Traditional BI Architecture

The engine of Alice is composed of Tensorflow, Python, NLTK, Linear SVM, React, Flask, Docker and NoSQL. The fuel is the tickets that have been filed in the past.