Natural language processing combined with artificial intelligence improves the speed and accuracy of interaction
COMPANY
UnitedHealthcare
THE CHALLENGE
Reduce the time members spend getting answers to simple questions.
THE OUTCOME
AiDA, a chatbot for both members and call center agents.
PRINCIPLES IN ACTION
THE STORY
One of the most time-consuming, and difficult, questions that call center agents get on a daily basis is verification of a member’s benefits and deductibles. Time consuming, because agents must query the right combination of information repeatedly, many times every day. Difficult, because agents have to sift through a large chunk of returned data to pull out the exact answer to a member’s question. Sometimes agents need to search for providers to get the full picture. And many times, the information returned is in dense medical jargon that agents have to translate for members.
The chatbot uses artificial intelligence and machine learning to deliver quick answers in plain language. Where appropriate, the interface includes graphical features such as charts and maps to display complex information. The team is building out a repertoire of interactions using conversational design principles.
Although still in its infancy, the chatbot already is fielding more than 400 queries a day with greater than 96% accuracy.