Wilmington’s most buzzy startup Next Glass
unveiled its app to the public today, beginning a journey to prove science is better than subjectivity in predicting taste.
Early indications are that Next Glass will be a hit—in a 24-hour beta launch earlier this month, the free app became the fourth most downloaded food or drink app in the App Store behind Starbucks and Dominos Pizza.
It hasn’t been hard to get media attention either—Next Glass made headlines in Business Insider, TechCrunch
and was an "app of the week"
. Next Glass executives took a national stage last month as one of five global startups to showcase at the Wall Street Journal’s
inaugural WSJDLive. It was named one of 10 Startups to Watch by the NCTA
at its annual Tech Awards.
And attention in communities around the nation came during Beer Census 2014. Next Glass sent two employees on a nationwide tour in a truck to buy every unique bottle of beer in 35 different cities, so they can be analyzed in a lab and loaded onto the app. The pair came home with more than 8,000.
The broad excitement around the app can be summed up by a comment Next Glass co-founder and CEO Kurt Taylor
made to me back in March.
“Everyone wants to discover something new but they don’t want to spend $30 not knowing if it’s going to be good.”
Investors have put more than $3 million into that big vision. It’s big because people care about more than just good-tasting wines and beers. Next Glass has a patent-pending process to measure the chemical make up of presumably any consumable substance, meaning future apps could help us find food or other beverages we might like.
“We are for the first time ever allowing people to shop and connect based on taste,” Taylor told me in an interview this week. “We’re creating this universal, objective language that allows people to know if they’re going to enjoy something before they purchase it. We're focused on wine and beer, but there are further opportunities outside those products.”
The Next Glass back story
Taylor is a UNC grad who worked briefly at a boutique investment bank in Charlotte before following in the footsteps of his entrepreneurial father to start a company. His dad, co-founder and key advisor is George Taylor, a serial entrepreneur who has been involved in eight startups, most recently a Wilmington automotive performance shop with plans to open locations around the nation.
The elder Taylor had a background in machine learning, in the mid-2000s using it to predict stocks. Netflix popularized the technology as a way of making recommendations, paving the way for Next Glass to happen.
The two men dreamt up the idea after selecting a highly-rated, sommelier-recommended wine at a Charlotte restaurant, and both disliking it. They lamented how difficult it was to trust a recommendation because every person’s taste is different. And they wondered if science, and an algorithm could help.
Machine learning only works when there are tons of data for it to make sense of. And having people test the wine would take too long. So they began to investigate a scientific process for measuring and cataloguing the makeup of each substance. They forged relationships with alcohol distributors (some even donated bottles) and hired a scientist from Washington D.C. to do the work.
The science and technology behind Next Glass
All that science happens in a lab at UNCW’s Center for Marine Science. It requires machines typically used to test blood and tiny syringes of wine or beer samples that are mixed with detergents and blasted with light to determine the transparency of the liquid. In that, tens of thousands of data points can be collected, including pH, glucose, alcohol content and for beer, carbonation. The younger Taylor compares it to “looking for a fingerprint.”
Machine learning technology looks for patterns in all that data, and Next Glass hired one of the best minds in the world, based in London, to build the prototype. So when you download the app and enter some of your favorite beers and wines, Next Glass can deliver recommendations based on the chemical compounds you like best.
The app also deploys augmented reality technology—hold your phone’s camera over a bottle’s label and see calories, nutritional details and a score that reveals the likelihood you might like the beverage. The algorithm improves the more beverages you enter and rate in the system.
During that first 24 hours, more than 20,000 people downloaded the app and many gave ratings. Both positive and negative feedback came from TechCrunch writer Sarah Buhr. She liked the big idea—that the technology could be used to measure other substances—and that she could find out caloric intake or use the app to find a cheaper version of an expensive wine she liked.
But the initial experience on the app was a bit clunky, requiring a user to scan through a bunch of random wines and beers to build a profile. Many tweaks were made to the app launching today.
There's also some competition in the space, apps that provide a notebook and camera so users can record their thoughts about a beverage they try. And Buhr named a handful of other beer and wine recommendation apps, admitting that Next Glass is the only to add intelligence based on science.
It's free, so how to make money?
Taylor won’t share much about the monetization strategy, besides that Next Glass will work with producers, restaurateurs and retailers to provide advertising or sponsorship opportunities, as well as data to help inform product development, menu or merchandising decisions. Back in March, Taylor called it “taste-based marketing.”
Other opportunities may come when Next Glass integrates location-aware features into the app. Today, users can only find out if a wine or beer is available in their states. Eventually, they can find specific stores to buy it in. (It’s been reported that former Harris Teeter chairman and CEO Tad Dickson is on the Next Glass board.)
But Taylor says that any money-making attempts will come second to providing the best recommendations for Next Glass users. Though some experiments will begin next year with Next Glass partners, revenue won't be a focus until 2016.
What is a focus, is building a new office in downtown Wilmington and a team to fill it. Taylor employs 15 today and expects to grow to 25 next year. The new office is expected to be complete by April 2015.
Marketing too. Channel partners will help Next Glass get customers in grocery stores and wine and beer shops to download and use the app. There will continue to be an online and social media push.
And Next Glass will continue to add to its library of wines and beers. More than 30,000 have been tested and computed today.
One concern has stuck with me about Next Glass—that the algorithm may be too limiting. I like to experiment, sometimes outside of my comfort zone. There's fun in the mystery of trying new things. How does Next Glass handle that?
Taylor has an answer. Next Glass doesn’t make recommendations based on type of wine, but on the compounds of notes.
Using himself as an example: “I never really considered myself much of a Pinot Noir fan, but after putting in a lot ratings, it started recommending pinots that I actually really liked."
It's all about the science. And I guess I better get rating.