Combining science and software to select your next glass
In a laboratory on the Wilmington, North Carolina, waterfront, a team of chemists is hard at work. Through the powers of liquid chromatography and ion trap mass spectrometry, the chemists are analyzing bottle after bottle of wine and beer. Their mission: to determine your next glass.
The startup company the chemists work for, in fact, is named Next Glass, and it is the brainchild of its 25-year-old founder and CEO, Kurt Taylor. Taylor is neither scientist nor wine expert, and he does not claim to be either. He is a man with an idea; one that he believes will revolutionize the wine and beer industry from a recommendation standpoint.
"What we’re working on is really groundbreaking stuff," Taylor said. "I think it does have the opportunity to change an industry, the way Pandora has changed music."
Comparisons between Next Glass and services like Pandora or Netflix are common, because the up-and-coming beverage "recommender" will employ the same machine learning technology that allows those companies to determine their users' preferences. The difference is that Next Glass will couple that technology with science.
Experience inspires app technology
Taylor’s company was inspired by a dinner that the young entrepreneur had with his father in the summer of 2012, when he was working as an investment banker in Charlotte, NC. The two were interested in ordering a bottle of wine and asked their waiter for a recommendation. After being assured that every sip of a certain bottle would be "more elegant than the last," they were presented with something that each of them utterly hated.
Frustrated by the subjective rating systems and vague "expert" opinions of taste that have long served as the only basis for classifying different wines and beers, Taylor began to hatch an idea. He wanted to develop a way to recommend beverages the same way Pandora recommends music—by taking a user’s favorite song or artist and developing an entire playlist of similar tunes based on that information.
The problem lay in how to go about classifying those beverages. Applications like Pandora and Netflix rely on classification algorithms, which in these cases are supported with data from experts listening to music and watching movies. That would not work for Next Glass.
"We realized that we couldn’t have people tasting wine and beer all day, because it would be a popular job, and we wouldn’t get much work done," Taylor said. "We had to figure out another way to define these bottles without any subjectivity or having any human interaction on the data collection side. That’s when we realized that wine and beer are different than music and movies in the sense that you can run them through a scientific instrument and learn exactly what compounds make them up."
So he left his job, obtained support from investors and his alma mater, the University of North Carolina, developed a team, and began to bring his idea to life.
Stocking the Genome Cellar
Enter the Next Glass team of chemists. Headed by Chief Science Officer Dr. Connell Cunningham, the crew Taylor has assembled is testing wines and beers at a furious pace, averaging 100 bottles per day. Their goal is to cover about 98% of the readily available U.S. market—more than 25,000 wines and 10,000 beers in total—by the time their application launches in June 2014.
The testing process begins with the extraction of a sample from a sealed bottle of wine or beer. The sample is analyzed using a mass spectrometer with Thermo Scientific™ Orbitrap technology, providing information that is mathematically converted into a spectrum of compounds—thousands for each beverage—that functions as that beer or wine’s "DNA."
Each bottle’s DNA is added to Next Glass’s Genome Cellar, the database from which the application's software can match it to users’ taste preferences based on other beverages they enjoy. A taste profile is created for each user so that they may discover new wines and beers that will align with their palate, or even to merge profiles with other users to identify a bottle that fits both of their tastes.
"This technology already exists and has been proven, as far as mass spectrometry is concerned," said Dr. Cunningham. "But what’s exciting is that we’re applying it in combination with these algorithms in a way that can really help people."
While Taylor believes that Next Glass will be a valuable tool for everyone involved in the wine and beer industries, from vineyards and breweries to liquor stores and supermarkets, he stresses that the consumer is meant to be the primary beneficiary. He hopes his application will become a tool that people can use to turn the task of purchasing new wines and beers--one that may have been daunting and frustrating in the past--into an enjoyable experience.
"We want to make it fun," Taylor said. "We want you to go into a store, interact with our application, and know with confidence that you're going to pick something that you're going to enjoy that night."
Next Glass will be available as an app for smartphones and tablets, as well as at kiosks in major supermarkets.
Finding new favorites
Taylor plans to release Next Glass to the public this summer. When he does, it will be with aspirations of changing an industry. It will also be with the confidence of knowing firsthand that his product works.
An unadventurous wine drinker for most of his adult life, Taylor primarily stuck to Cabernets that he knew well. "Interestingly," he said, "on the opposite end of the spectrum from a Cab would be a Pinot Noir. Through our software I've already found three or four Pinots that have a lot of the same chemical notes, and they've become some of my new favorites."
Through his unprecedented combination of science and software, Taylor has already found his next glass. Soon, Next Glass will help you find yours.
Learn how chemists utilize liquid chromatography and ion trap mass spectrometry to determine the "DNA" of wine and beer.