One of Australia’s largest superannuation funds with millions of members was looking for help to optimise their chatbot capability. They wanted to improve their customer experience in the chat by ensuring readily available information was provided by the chatbot, thereby freeing up people to focus on solving complex customer queries.

The Super fund engaged with Servian to use our AI knowledge to develop a chatbot capable of intelligently dealing with FAQ type queries. The chatbot was developed in Google Cloud Dialogflow chatbot on GCP which utilises Natural Language Understanding to interpret a users question and respond accordingly. The chatbot can respond to over 100 different intents with an accuracy of 90%. This means that 70% of user queries are dealt with in a way that is intelligent and accurate, leaving the customer feeling satisfied and passing only the most complex queries to the live agents.

As well as creating an intelligent, human feeling, chatbot – Servian also designed a framework to automate populating and training the bot. The framework includes running automated test cases through an interface to measure accuracy. The use of the framework means that the training data for each intent, and the intent responses, can be stored in an environment outside the DialogFlow console where it can be versioned and controlled.