The Digital Twin That's Transforming the Way People Shop

Shoppers like to support their local stores, but retailers don’t always make it easy.

The benefits of browsing and interacting in person are often filled with frustration:

•   A customer arrives only to discover the clothing item they wanted is out of stock.

•   Their second-choice item in their size doesn’t quite fit.

•   They make it to the checkout only to discover a long line of customers waiting to pay.

•   They finally get to the register and the barcode won’t scan.

As the cashier calls for the manager, the customer might wonder why they left their home. But what if retailers could take all the strengths of internet shopping, and bring them to the in-store retail experience? They’d create a seamless retail journey, pleasing consumers while boosting their profits. Welcome to frictionless retail.

Artificial intelligence is solving these problems and unlocking a new era in retail.

It’s made possible, in part, by Intel’s OpenVINO framework.

How does it work?

AI algorithms can analyze real-time data to help personalize customer interactions, provide efficient inventory management and streamline checkout processes. Until recently, running deep learning models required costly and advanced computing equipment. This meant that dynamic AI capabilities were available only to organizations with significant resources and infrastructure, such as data centers and manufacturing plants. To increase access to frictionless, personalized experiences, Intel’s OpenVINO framework fine-tunes and optimizes powerful deep learning models so they can run efficiently on various new and existing devices right in the store.

What does it mean for retailers?

This allows cameras, sensors and other in-store devices to “think” and make data-driven decisions on the spot. “That could be on a kiosk in a retail store,” says Adam Burns, vice president of Intel’s network and edge group and general manager of OpenVINO and developer tools. “Or it could be on the back-of-house servers in the retail store doing inventory management.”

What does it mean for my privacy?

It’s a positive step for customer privacy because data is being collected and analyzed locally rather than transmitted to a remote cloud data center. OpenVINO allows for real-time analysis, meaning that data can be analyzed and then immediately discarded if it’s not needed for long-term storage. This lowers the risks associated with data breaches.

One of the technology’s most eye-catching use cases so far is the digital twin, which makes clothes shopping easier.

Intel’s partners can leverage OpenVINO technology to turn the frustrating process of trying on clothes in a fitting room into an automated experience that tells you exactly which items are right for you.

Technology platform Fit:match, for example, provides stores with high-tech fitting rooms where an array of sensors analyze the shopper’s body shape to create a digital twin. Its patented shape-match algorithm then intelligently suggests various well-fitting clothing options for the customer. This fusion of augmented reality, computer vision and artificial intelligence takes just 30 seconds. Customers can then receive an email showing a selection of clothes that are likely to suit them. They can either find and buy those items in the store—knowing they will fit—or buy them with confidence online. “They’re walking out of that store satisfied,” Burns says. “They feel comfortable that their data is only resident within that store and isn’t going to compromise their own privacy.” There’s an environmental benefit as well as a profit motive, Burns says. “When you can put the customer in something they know they’re going to like, there’s less waste from returns … [and stores] don’t have to do the extra processing of going through returns. If you’re talking about online retail, you’re not shipping things that are then just shipped back.”

It’s just the beginning.

Here are some of the other ways the shopping experience is being transformed with OpenVINO technology.

Lunch is served. (They knew you were coming.)

When you have cameras monitoring the fast-food parking lot and how many customers are in line inside, algorithms can make sure the kitchen has the right amount of food in progress to keep the line moving. “At my peak times during the lunch hour rush … I want to make sure I’m producing the right amount of food for the right customers,” says Burns. “I’m not wasting anything, but I’m also getting people through as fast as possible.”

Self-service kiosks. White-glove service.

“Would you like a muffin with that, sir?” The self-service terminal already knows the answer. If a coffee chain can recognize you coming in the door, it can also learn your routine and preferences. For example, on Friday, you might typically let your hair down a little. After four days of an espresso and a banana, an automated kiosk can anticipate that you’ll pivot to a bakery item and a latte to round off the week.

Out of stock is out of style.

Sure, store employees can spend their time monitoring shelves and stock levels. But aren’t they better used to provide the level of customer service that sets in-store shopping apart from online? “You can include a number of sensors on the shelves to make them intelligent by weight or by visual, understand which products are being emptied out and so on,” says Gustavo Reyna, segment market strategist at Intel.

Self-checkout that won’t make you tear your hair out.

Self-checkout can be a headache, but if the checkout can “see,” there’s less room for error. Instead of scrolling through a list of store items to find the exact brand of apple you’ve got in your basket, the checkout can just know. It can also thwart those who try to pass off more expensive items as cheaper ones. “The loss of revenue from objects that are taken out of the store or checked out erroneously … ends up being a multi-billion-dollar shrinkage for retailers,” Reyna says.

The more data, the more insights.

When abundant data is processed efficiently, it gives retailers deep insights into their operations, enabling improvements that can optimize and differentiate their businesses, enhance customer experiences, foster brand loyalty and ultimately increase revenues. “As you collect all of that data, then you begin to create models where you can anticipate certain trends with customers, certain times of the day, certain items that are sold more than others,” Reyna says. “So all of that data becomes very useful, very powerful insight from the retailer into how their operation is working.”

‍