An expert store associate doesn’t just answer a shopper’s questions, he digs for more details in order to come up with the best product suggestions and advice he can give her. Digital agency Fluid Inc. is working to bring that capability to a shopping app, leveraging the technology of IBM Corp.’s Jeopardy-winning cognitive science computer Watson to do so, says Fluid CEO Kent Deverell.
Watson is an computer that builds upon advances in artificial intelligence so that it can engage in dialogue with people, learning over time how to improve the relevance of its answers and its conversational skills, according to IBM. In 2011, Watson beat the top two human competitors in the history of the trivia game show Jeopardy during a special challenge edition of the show, which IBM says it considered the final test of the machine’s learning abilities.
Fluid plans to launch a prototype of the Watson-powered app, called the Fluid Expert Personal Shopper, next year, Deverell says. It is working with outdoors gear and apparel retailer The North Face to develop a test version of the app, which will most likely launch for the iPad, he says.
The North Face is one of several brands owned by VF Corp., which is No. 113 in the Internet Retailer Top 500 Guide. It did not immediately respond to a request for comments about the app.
With the app, a North Face shopper, for instance, could speak or type into the iPad a question such as, “I’m going on an expedition to Patagonia. What gear should I bring?” Then Watson, via the app, asks questions to refine what she needs, perhaps to find when she is going or how many miles she plans to trek per day, Deverell says. If the shopper says her expedition is planned for the winter, for example, Watson may suggest she look for a backpack with ABS, an airbag-based avalanche rescue system used by climbers.
“You would never have that conversation with Google,” Deverell says. Unlike Google Inc.’s search engine or Apple Inc.’s voice search service Siri, which both mine online data using keywords, Watson also takes into consideration the context of words to determine what answers are most relevant. “You can think of it like Siri on steroids,” he says. For instance, Watson would know that a shopper seeking ABS gear for a trip to Patagonia in the winter is seeking avalanche technology specific to climbing. Whereas, if that shopper searched for ABS gear on Google, 90% of the results would refer to anti-lock braking systems for cars—the most common meaning for the three-letter phrase on the search engine, Deverell says.
IBM hosts the Watson technology on Internet servers, which Fluid taps into to facilitate conversations between the computer and customers through the Fluid Expert Personal Shopper. Fluid also programs retailer-specific rules and features into the app. For instance, a retailer might set rules for which products the app can recommend, such as to avoid promoting out-of-stock products.
The North Face is contributing to the app prototype its product catalog, call center logs and ratings and reviews information to help Watson devise product suggestions and further questions for a shopper, Deverell says. Fluid writes software programs that train Watson on how to interpret and use the retailer’s data in its conversations with customers.
Watson already is connected to many other publicly available data sources that it uses to come up with shopping advice, including the entire contents of the free, user-generated encyclopedia Wikipedia, Deverall says. Some of those sources in particular, such as online consumer reports and magazine articles about products, will help provide more shopping-specific data to the app as it develops, he says. Eventually, the Fluid Expert Personal Shopper may also connect to information about an individual shopper, such as her social media profiles and purchasing history, which will allow it to fine-tune its advice based on her preferences.
The Fluid Expert Personal Shopper is launching first on an iPad because that device can demonstrate both voice-based and text-based Q&A sessions, Deverell says. The agency plans to make the app available later on any type of device, including as a web site feature accessible on a laptop or desktop computer. However, Deverell says the technology seems most promising for smartphone shoppers.
“There’s a huge trend towards smartphones, and this interface is perfect for that,” he says. “Shopping is hard on the small screen, it’s hard to browse and filter, so being able to actually speak to Fluid as your personal shopper will be very powerful.”
In the future, he also sees the app becoming a tool to help the workers it was originally designed to mimic—store associates, by providing them with the additional expertise of Watson via a smartphone or tablet in the store, he says.
Pricing for the app is not yet set.