Learned it through the grapevine
Houston tech company uses AI to help customers select wine
Amy Gross was in Napa Valley 10 years ago when she noticed her taste in wine, a preference for subdued, earthy flavors, was different than her husband’s affinity for bolder, peppery wine. And both of their preferences were different from the two other couples on the trip.
The six people were similar ages and lived on the same street. Two of them drove the same vehicle, a Toyota Sequoia. They agreed on a lot of other things — except for the nuances of wine.
“It struck me how we had similar tastes in so many things, but when it came to flavors we had different opinions,” Gross said. “And I thought there had to be something to that.”
That’s when she got the idea for VineSleuth, a machine learning platform that asks customers about their taste preferences, such as how sweet they like wine or how hoppy they like beer, and then recommends beverages available at grocery stores, bars or restaurants. VineSleuth, however, has a different business model than other wine apps, such as Vivino, in which consumers who download the app can take a picture of wine bottles to learn about the wine, rate it and buy the bottle to be shipped to their home.
VineSleuth’s customers are the stores, bars and res
taurants that offer the app to help its customers narrow their selection of beer and wine. This business-to-business approach was a pivot made by Gross many years into her entrepreneurial journey, and it recently helped VineSleuth raise $570,000 and launch its product at a wine tasting room in New York.
The idea for VineSleuth started after that trip to Napa Valley in 2009. Gross started a wine blog, where she learned about the intricacies of making, tasting and selling wine. She later began judging wine competitions at the Houston Livestock Show and Rodeo, California State Fair and other international events.
She worked with a Cornell University professor who specializes in evaluating wine flavors and chemistry, applied mathematicians and winemakers to analyze wines based on flavors, such as how strong it tastes of red fruits or how much sugar would be perceived by drinkers. This collaboration created the company’s methodology for analyzing wines.
Every wine listed on VineSleuth is analyzed by a tasting panel, and that data is put into a patented algorithm. When customers input their flavor preferences, the machine learning technology makes the proper recommendations.
She launched the company in 2014 with a consumer focus. The app’s users rated wines with essentially a thumbs up or thumbs down. But Gross realized it was difficult for consumers to know where they could buy those wines.
Practical information
For instance, a shopper at Kroger or H-E-B wants to know about the selection at that grocery store. It doesn’t help to know about wines that can’t be purchased at that moment. It’s the same problem at restaurants and wine tasting rooms.
“What makes our system work so well is our data,” Gross said. “It’s the way we analyze those wines plus the machine learning. So to keep the integrity of our data, I knew that we needed to start with a smaller subset of wines in a place where people could actually use that information.”
Gross created a business-tobusiness company aimed at helping retailers and restaurants better guide customers in selecting a beer or wine they’ll enjoy. The product was launched in September at Fox Run Vineyards in the Finger Lakes region of New York, where Cornell is located.
Fox Run Vineyards has a tasting bar where customers typically look at a paper wine list and select what they’d like to try. Now, employees at the tasting bar help customers use a tablet to select characteristics they like in a wine and how much they like it, sliding their preference for acidity, minerality, sweetness, oak flavor or fruit flavor on a scale ranging from none to a lot.
It intentionally does not ask if customers prefer red or white wine because that could prevent them from trying something they might like. One customer, for instance, identified himself as a red wine drinker. The test, however, recommended a dry Riesling.
“He left buying six bottles of dry Riesling,” said Scott Osborn, president and co-owner of Fox Run Vineyards. “He said, ‘I would have never tried the dry Riesling because all I wanted to do was taste red.’ So it (VineSleuth) has opened up people’s minds to try wines they’ve never tried before.”
Osborn was intrigued by the app because he thought it could boost customer engagement. So far, the tablets have prompted conversations with customers that last five to 10 minutes longer, though that can cause long lines on the weekends when the tasting room is full.
A second winery in the Finger Lakes region is poised to start using the VineSleuth app soon. VineSleuth also will be unveiled next month at a craft beer bar in Spring called Thistle Draftshop. A few restaurants are slated to start using VineSleuth in the coming months.
Gross is using Osborn’s feedback to make adjustments to the user experience and to make the app more intuitive, so it can be used in both fast-paced environments, such as a busy bar, or more leisurely ones, such as a fine-dining restaurant.
Upsell
Ultimately, she thinks VineSleuth could help wine makers, restaurants and grocery stores entice their customers to spend more money. Casual wine drinkers, which Gross said makes up the majority of wine drinkers, might be comfortable spending $7 on a bottle of wine, but hesitant to try a more expensive bottle because they are worried they might spend more money for something they don’t like.
But with VineSleuth, Gross said, maybe they could get the confidence to try a slightly more expensive bottle.
“If you can get them up to $9 or $10, that’s going to make a much bigger difference than appealing to the high-end wine drinker that there are fewer of,” Gross said, “and you’re probably not going to expand their spend.”
andrea.leinfelder@chron.com twitter.com/a_leinfelder