IBM’s Watson has a summer job working at Macy’s
The Jeopardy-winning supercomputer is now venturing further into retail, in its first application inside a brick-and-mortar retailer.
IBM’s Watson supercomputer is now a shopping assistant at Macy’s.
Since it vanquished several human “Jeopardy” opponents in 2011, the famed supercomputer has been making its living by helping medical researchers, marketers, oil companies, advertisers and other professions. And, this summer, it is powering Macy’s On-Call, a mobile web application that is the first use of Watson inside a brick-and-mortar store.
At Macy’s, Watson’s remote intelligence is delivered via location-based mobile engagement platform Satisfi. An in-store user accesses On-Call through a mobile web browser, asking natural language questions about products, services and facilities.
For instance, the user might type a natural language request relating to a store department, brand or product, such as: “Where are men’s shoes?” Or he might ask for a brand plus a product, like “Alfani shoes.”
Watson will return brief product info and the product’s location in the store, as well as any relevant Macy’s human services or nearby departments that might be useful, such as [email protected]’s personal shoppers.
IBM Watson Program Manager Jonas Nwuke emailed me that On-Call is “more than just a product map,” since it is designed to provide customers with “context in its response, [interacting] as if they were having a conversation with another person.”
“For example,” he added, “it will provide contextual information about what other departments are near [a specific] product,” such as: “The ladies’ shoes are on the second floor, across from the cosmetics department and next to jewelry.”
Macy’s own mobile app lets a user order out-of-stock items or scan products for price. The retailer’s Group VP of Digital Media Strategy, Serena Potter, told me via email that On-Call “is the first time Macy’s is providing a mobile tool that serves as an in-store companion,” designed to help customers navigate specific stores and to learn over time.
This first phase of the IBM/Macy’s pilot project focuses on providing a discovery tool at 10 selected Macy’s locations, and the companies said they will evaluate the project — which ends in late fall — to see if they should move to a next phase, which IBM said might include “full cognitive dialog capabilities.”
Of the 10 locations, five are characterized as base learning stores, where customers use the app as self-service. The other five stores have an enhanced layer of sales associate support to complement On-Call, such as a request for a face-to-face engagement with a Macy’s specialist associate.
IBM’s Nwuke told me that Macy’s conducted an extensive survey of the 10 pilot stores to “map out the location of departments, products by brand and category and in-store services,” all of which was conveyed to Watson.
“From this data corpus,” he said, “Watson is not only able to answer questions shoppers ask of it, relevant to that location, but the tool learns every day as it interacts with consumers.”
Base learning locations are: Macy’s Montgomery in Bethesda, MD; Macy’s Woodbridge Center in Woodbridge, NJ; Macy’s Clackamas Town Center in Portland, OR; Macy’s Santa Anita, in Arcadia, CA; Macy’s Miami International, in Miami. The stores with sales associate support for On-Call: Macy’s Short Hills, in Short Hill, NJ; Macy’s Mall of Georgia, in Buford, GA; Macy’s Lenox Square, in Atlanta; Macy’s Aventura, in North Miami and Macy’s Roosevelt Field, in Garden City, NY.
Watson’s other retailing gigs
While this is Watson’s first foray into brick-and-mortar, it’s not its first gig in retailing.
Late last year, IBM launched with outdoor retailer The North Face an interactive online shopping experience designed for natural conversational interaction via computer.
A user might type, “I need a jacket for a Vermont ski trip” into the Watson-enhanced recommendation engine, powered by digital marketing agency Fluid’s Expert Personal Shopper. The system then asks the kinds of questions a sales associate might, like the trip’s location, the temperature and the gender of the wearer, and it comes up with recommendations. As with Macy’s On-Call, it is designed to improve over time.
This past May, Watson was also utilized for 1-800-FLOWERS.COM GWYN (Gifts When You Need), another Fluid XPS solution that IBM describes as “a cognitive-powered gift concierge” that similarly has a natural language interface for a recommendation engine that learns from experience.
Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.