Software For Hardware: How Artificial Intelligence Is Helping Lowe’s Customers – AI| AI
magine two typical problems facing a home improvement project: a carpenter trying to find just the right nail quickly in a big-box store and getting more frustrated with every passing minute; his customer trying to picture just how an outdoor deck will look after a renovation. Lowe’s is turning to AI technologies to give customers—and its employees—a big helping hand.
The venerable home improvement retailer probably isn’t the first name that comes to mind when you think of artificial intelligence and robotics. Yet it’s fitting that a chain selling hammers and saws would clearly see AI technologies for what they are—tools to get a job done. Welcome to a world where software meets the hardware at a home improvement brand.
SEE MORE FROM PART ONE OF A SPECIAL SERIES ON AI
Gihad Jawhar, vice president of digital development at Lowe’s and the guiding hand behind the company’s AI efforts, describes the big picture behind the brand’s deployment of various technologies—computer vision, machine learning and natural language processing (NLP)—to solve customers’ problems and guide them seamlessly and quickly to the right products. “AI happens to be the right tool for us to do these things,” he says. “We never set out to leverage AI. The technology doesn’t drive the solution. The problem drives the solution—and AI technology is a way to solve it.”
Lowe’s is hardly alone—a recent Forbes Insights survey found that 50% of retail executives chose AI as the most important technology for the future of their companies.
AI As A Helping Hand In Customer Service
Lowe’s has experimented with a number of AI-driven projects, and it’s set to deploy new solutions. Its LoweBot in-store autonomous robot, launched as a one-year pilot program in 2017, was able to answer basic questions and navigate customers and employees through a store. It kept track of inventory in real time and detected sales patterns that could guide business decisions. Now the company is taking some of the technologies from that experiment—computer vision, for example—to spot holes in inventory—literally empty space on store shelves. And it has just released a text-based customer service platform that will anticipate what a customer wants.
All of these things enable Lowe’s employees to devote their attention to more complex, human problems, says Jawhar, like lending their expertise to a customer’s particular project, perhaps a troublesome leaky roof.
The LoweBot and many of the retailer’s other AI technologies emerged from Lowe’s Innovation Labs, which is also working on technologies like augmented/virtual reality visualization and an app-based VR store (the first of its kind), where customers can see how products or projects will look in their homes. An example of the company’s focus on visualization in retail is its Holoroom Test Drive, a fully immersive VR experience that allows customers to test power equipment in a true-to-life virtual way. The technology won the Auggie award for Best Enterprise Solution in 2018.
“We utilize our stores as living labs to rapidly test our prototypes and gather real-world feedback on the new experiences we’re delivering to demystify home improvement,” says Josh Shabtai, director of Labs Productions & Operations at Lowe’s Innovation Labs.
Texting and search get smart
The innovative experiences don’t stop there. Jawhar and his team are also building a customer-service experience around texting, which is the dominant communication platform for most Lowe’s shoppers and the ideal forum to deploy natural language processing and machine learning.
The solution began as a simple texting option and expanded recently to Apple Business Chat. “We found that customers were asking for very similar things during the customer service experience,” Jawhar says. “They were just asking for it in 10 or 15 different ways, depending on where they were in a project, where they were located, the type of words that they used, and how emotional or non-emotional they were about the query. In testing, we were able to identify accurately about 70% of customer intents and improve our algorithm from there.”
Lowe’s is also using AI technology to sharpen its website search engine, so it knows that “leaky roof,” for example, actually relates to products such as shingles or siliconized reflective roof coatings. “We can input AI-driven insights, results or signals captured from searches to get the customer to the right place,” Jawhar says, “but to do that, we’ve got to understand the human language, and we’re finding people on our website search engine are starting to ask questions as if it were a human.”
AI eyes the shelves
Computer vision is another technology being deployed by the retailer. Small cameras, now mounted at the top of shelves in certain high-touch departments, stream real-time information on shelf-stock levels. For example, the technology detects when a hole appears in the light bulb section, and the system then sends a real-time notification to the store’s devices so staff can quickly head to the stock room and replace the item. “The system can determine the specific item—its precise shelf location and whether or not there is additional stock in the store—all before the alert reaches the associate,” says Kevin Seidehamel, director of market concept and development at Lowe’s.
Artificial intelligence is just beginning to reshape retail. Jawhar says that despite the progress Lowe’s has made to date, its AI efforts are still in the early innings. This is true across the board, as Forbes Insights research reveals that just 3% of retail companies have fully deployed artificial intelligence, while 26% report that it has already become a significant part of their business. A majority of retail companies (64%) consider themselves slow at implementing the technology.
“We found that customers were asking for very similar things during the customer service experience. They were just asking for it in 10 or 15 different ways.”
How did Lowe’s start to build and deploy its AI? “We did a test run similar to a minimum viable product,” Jawhar says. He and his team used the strong training data from their test runs to then feed the algorithms continuously and make them more and more intelligent. “You end up at a point where you’ve got a machine that’s managing the more routine tasks, freeing up humans to do what we excel at. This is a huge point of progress and efficiency,” he says. Once you get to a success rate of 95% or so internally, he notes, you can do it for a small segment of customers. Then you measure the satisfaction of those customers. “It’s the way you accomplish many things in business,” he adds. “Make it small enough that failure and the financials around a failure are an acceptable model, and then just bust tail to make it as successful as possible.”
Click here to receive notifications about future updates on artificial intelligence and other relevant topics from Forbes and/or Intel.
CREDITS: Lowe’s Innovation Labs; Forbes Insights study