Felix Laboy, CEO of WayBlazer, today provided Etopia News with the following responses to its recently-submitted questions regarding the operations of “Connie the Concierge,” a small, semi-humanoid robot that channels the power of IBM Watson into answering questions about travel and tourism destinations:
How is the project working?
The pilot is a joint collaboration between
Hilton and IBM Watson together with WayBlazer. Connie enables guests at
the Hilton McLean in Virginia, to interact with the first Watson-enabled robot
concierge, to learn more about in-hotel offerings and local attractions.
When guests ask questions in natural language
about restaurants, attractions and things to do “off-property” (Local Area
Knowledge) WayBlazer is able to make local recommendations in the form of
visual tiles including images and insights about the POI.
How much is it being used?
WayBlazer focuses on the local area knowledge
component of Connie’s recommendations. The technology combines
WayBlazer’s extensive travel domain knowledge with IBM Watson's cognitive
computing technology platform to enable Connie to greet guests upon arrival and
to answer questions about hotel amenities, services and hours of operation.
Guests access local area knowledge data daily with a large focus of questions
on restaurants in the local area, as well as places to grab a drink and
attractions like movie theaters, parks, and museums.
How people are reacting to it?
The WayBlazer team is not on the ground in McLean but from the data we can see that guests are engaged with the local
area knowledge piece that WayBlazer provides evidenced by the number of
queries, clicks into recommendations cards and clicks for directions.
Plans for the future
WayBlazer is excited to see where Hilton and
IBM take this project next and will be ready to support with our natural
language search and local area knowledge as well as other technologies where
there is a fit.
What has WayBlazer learned?
Our training process has allowed us to learn
from guest queries and continually improve the product to be more accurate in
understanding the questions and providing a relevant response. We are currently
at an accuracy of 90%.
