6 retail leaders innovating the customer experience with AI

Five years ago — well before artificial intelligence came to dominate headlines — Walmart embarked on a journey to build a platform upon which it could layer dozens of services for customers and employees alike.

Starting with an open-source architecture, Walmart's tech team trained its AI to understand retail concepts and Walmart-specific brands and terminology to power everything from customer-care chatbots, to a text message-based shopping tool, to the "Ask Sam" digital assistant for store associates.

"We're actually thinking about things as foundational capabilities, and laying that foundation, which actually allows us to be really fast," said Desirée Gosby, vice president for emerging tech for the company's Global Tech division.

Since joining the team three years ago, the MIT-trained mathematician and computer scientist is spearheading the development of AI services that make it quicker and easier for shoppers to find the product or assistance they need.

The intellectual challenge of inventing new custom tech solutions for the world's largest retailer is certainly exciting, but for Gosby and her team, the focus is first and foremost on delivering relevant, accurate, helpful results for the company's 230 million customers. 

Indeed, the two elements go hand in hand: "If you're looking at problems from a customer's perspective, you're going to find some really interesting research questions to solve," she said.

As generative AI projects gather steam and mesh with other services like augmented reality, new tools will soon enable customers to plan their kid's birthday with a custom list of party supplies, or find housewares that match their existing decor.

"It's still early days," she says of generative AI. "This is a new tool, but it's giving us an opportunity to solve customer problems in new ways."

With Little Caesars' AI-powered Caesar Vision tools and the brand's mobile app, a hungry dad can order a large pepperoni pizza (and a side of crazy bread, of course), pick it up from a nearby restaurant's heated Pizza Portal, and be on his way home in a minute or less.

The company's ability to serve up fresh pizza on demand is the result of what Chief Information Officer Anita Klopfenstein calls "pizza forecasting."

The machine-learning model analyzes five prior years of restaurant-level data to predict the exact amount of prepped ingredients and finished pizzas that each US location should have ready to go at any hour of every day.

Klopfenstein says distilling the complexity of the world's third-largest pizza chain into something so seemingly simple for employees and customers is critical to the brand's growth: "If it's difficult to use, they're not coming back."

Keyvan Mohajer launched the voice-recognition platform SoundHound, then called Melodis, in a Stanford dorm room in 2005.

The doctoral student knew "conversations with robots" were the future of customer service. But he couldn't get any investors to fund technology that would take a decade or more to perfect.

"In those days, they wanted faster outcomes," Mohajer told Insider.

So the CEO and his cofounders "went back to the dorm room" and came up with an idea that would change the trajectory of the business.

"We can build a humming engine so you can hear a song and we can tell you what it is," he said.

SoundHound was born. For the next several years, the music-recognition app helped fund "stealth" voice AI research at the firm.

"We knew that voice is going to be everywhere and conversational AI is going to be a future," he said.

After years of research, SoundHound's voice AI is changing how thousands of consumers order food at hundreds of restaurants across the US, including White Castle. The company's voice AI bots take phone and drive-thru orders.

White Castle is rolling out SoundHound's voice bots to 100 drive-thru locations. Other restaurants and chains adopting SoundHound's bots include Beef O'Brady's and Kneaders bakeries.

SoundHound said its tech is entirely automated and doesn't require human interaction. The bots are fast and tackle order-taking tasks that "restaurant staff don't want to do," Mohajer said.

Oracle, Toast, Square, and ChowNow, online ordering and point-of-sale tech companies, offer SoundHound's voice AI as an option for their restaurant clients.

Rivals in the space include Presto, ConverseNow, and OpenCity, whose voice AI bots are used by Hardee's, Panera Bread, Domino's, and Wingstop. Mohajer says he likes the competition. It proves the tech has gone from stealth to mainstream adoption.

"This is the year that it's finally happening," he said.

As a vegan, Shawn Stover, vice president of SmartHome Solutions at GE Appliances, has seen firsthand the benefits of having AI tools that can help create recipes within his dietary parameters, including a recipe for a cashew cream sauce. 

Stover is one of the leaders who helped design two generative AI tools for GE Appliance's SmartHQ app, including the SmartHQ Assistant and a recipe-generating tool called Flavorly AI. The tools were developed using Google Cloud.

Flavorly AI creates unique recipes for users based on items they already have in their kitchens. When using Flavorly AI, users can select a breakfast, lunch, appetizer, dinner, or dessert option for generating a recipe. They can then add up to 10 ingredients, along with any dietary specifications. From those instructions, Flavorly AI will generate personalized recipes that users can favorite or save for later on the app. 

Stover said the benefits of using Flavorly AI range from providing people with cooking inspiration to eliminating food waste by generating recipes using items in people's fridge or pantry that may be close to expiring.

Consumers who have GE Appliances in their homes can turn to the SmartHQ Assistant to ask questions about the appliances in their homes — questions you would typically find in appliance care manuals or by Googling, Stover noted. The assistant will tailor its answers to the consumers, based on the appliances connected to their SmartHQ account.

"Let's face it, none of us probably know where our user manual is for our refrigerator or dishwasher. We put it in a drawer or we threw it away or it's in a filing cabinet." Stover said. The SmartHQ Assistant "really tailors its responses for its consumers. They can ask it things like, 'What is convection bake?' and 'What filter goes in my refrigerator?'."

Stover sees generative AI continuing to change the appliance industry, by making appliances more personalized for users. An empty-nester may use a dishwasher or fridge very differently from a family of five, Stover said. He also thinks there's an opportunity for generative AI in appliances to cross over into the wellness space.

"I think there are teachable moments with generative AI. You can get younger members of the family involved in things like cooking, baking, cleaning, teaching them to do their own laundry," Stover said. "I think these are all opportunities for us to just teach life skills, and help people live better lives."

Particl, which launched in 2020, gathers publicly available data and uses an algorithm to predict sales. Brands use the tool to see how they measure up to their competitors, determine prices, or research new product launches. Currently, the company is tracking about 500 million products across the web, Wilson said.

"It's a busy, loud world, and Particl just tries to synthesize that," Wilson said. "Our customers can know the immediate impact on their business and it's backed by what's actually going on rather than just someone's opinion."

Particl's clients include apparel and shoe companies True Religion, Koio, Cupshe, and most recently, Kim Kardashian's shapewear brand Skims.

This year, Wilson has seen a growing appetite for companies to "talk with their own data" to better assess how products are performing. "The early power that's come out of this AI shift is the ability to have a conversation with data," he said. "Before you used to have to be quite skilled and trained in order to get those answers."

Clients can integrate their own sales data with Particl's data. Then using AI, Particl can predict how reducing a price could increase sales or identify a color or pattern that competitors have. For example, Particl suggested that one brand launch a pant based on demand it saw.

"Chubbies has been one of our best customers since we started," Wilson said. "Chubbies created pants based on bestselling pants from Lululemon and other competitors and have performed extremely well."

On the horizon, Wilson sees AI technology changing the way customers find and buy products. "I don't think there's quite a great implementation yet," he said, but he believes there's a large market for conversational shopping.

JJ  Zhuang sees lots of applications for artificial intelligence at Instacart, from keeping down fulfillment costs to making substitutions when something a customer orders is out of stock at a store.

But one use of AI that the former Microsoft distinguished engineer has spearheaded at Instacart helps customers tackle a more basic question: "What should I make for dinner tonight?"

Instead of bouncing between recipe websites that pop up on Google or watching videos on YouTube for inspiration, customers can use Ask Instacart to pose questions directly to AI and get an answer "the way you might ask a family member or a roommate or a food expert," Zhuang told Insider.

Those questions can be specific, such as "What kind of side dishes should I serve with lamb chops?" or "What are dairy-free snacks for kids?" But Ask Instacart can also tackle more open-ended queries, Zhuang said, such as: "What should I make for dinner if I have kale and cheddar in my fridge?" The function can also take into account factors such as customers' budgets or dietary restrictions.

Ask Instacart answers these questions with products that users can then order through Instacart, Zhuang said in a May blog post.

By using AI in this way, Instacart is changing the grocery shopping experience for its customers "from one that is utilitarian and keyword search-based to one that is inspirational and uses natural language," he said. 

"Ask Instacart introduces AI-supercharged product search in a way that we believe all commerce search is headed," Zhuang added.