Tastry, a San Luis Obispo, California-based sensory sciences company, has a bold claim.
“We say we taught a computer how to taste,” says Katerina Axelsson, CEO and co-founder. “But really we figured out how to look at the flavor matrix and relate it to human palate perception.”
The results make Tastry poised to forever change how wine is made, purchased, and perceived.
“Looking at the chemistry”
Katerina Axelsson, CEO and co-founder, Tastry
Tastry’s corporate description is simple. It states that the company “uses advanced chemistry, machine learning, and artificial intelligence to match consumers to products they will love.”
Tastry does this by using a robust database. Part of that database is comprised of analyses of tens of thousands of wine samples using “advanced GCMS-like technology.” This looks at volatile and non-volatile chemical compounds, ion counts, spectral information, and other data.
The next part is a massive database of consumer wine perception. The company asks people how much they like or dislike wines that Tastry has analyzed.
The final part is asking consumers questions that have to do with palate preference. Whether someone likes, say, Pellegrino, dark chocolate, or licorice. Using machine learning and artificial intelligence algorithms, Tastry then uses all of this information to reliably predict whether or not an individual will like a specific wine.
“What we do is we can understand how much consumers would like a sensory-based product, like wine, with 93% accuracy just by looking at the chemistry,” Axelsson says.
How many questions does it take to do so? “It takes about 20 questions for a red wine or 18 to 20 questions for a white wine for us to understand your palate with no prior data on you,” Axelsson says.
“One big, giant virtual focus group”
The potential applications for Tastry’s software are vast in the wine industry and beyond. (Tastry has largely focused on the wine industry to date.) They include product development, winemaking decisions, sales, and marketing. There are consumer applications as well.
Tastry can provide wine chemistry information, analyses of competitor wines, and the palate preferences of a winery’s customers. These data can be used for everything from maintaining vintage to vintage consistency to making a wine more consumer-appealing. It can even be used to identify specific oak profiles and find coopers with similar profiles, a process that often takes winemakers years of trial and error.
Via the company’s CompuBlend® functionality, Tastry can suggest blends that will have high consumer favorability. Winemakers can run simulations of different blends and make decisions. They can also see how their wines, or virtual blends, compare with their competitors.
Tastry also has a software add-on that integrates into e-commerce sites. It accumulates data about consumer preferences and recommends wine. Tastry can even tell where a particular product will be popular. The consumer-facing end of the software captures zip code data.
“To sum up what Tastry does, it’s kind of like one big, giant virtual focus group that you can run simulations against for what consumers are going to like,” Axelsson says.
“A complete game changer”
Cameron Hughes, founder Intergalactic Wine Company
Cameron Hughes is the founder of Intergalactic Wine Company, a global brands and private label solutions platform. He has spent a lifetime offering quality wine at affordable prices at Cameron Hughes Wines, De Négoce, and elsewhere.
“I think this is a complete game changer,” Hughes says.
When Hughes met with Tastry “the light bulb exploded over my head.” He put Tastry’s software add-on onto a company website to better understand his customers’ palates. This enabled customers to find wines that would be appealing to them, with feedback that it was “very accurate.”
Hughes has also used Tastry to understand the potential appeal of a wine and in blending. “I put in a bunch of different sub $10-a-gallon Central Coast Pinot Noir, and dammit if this thing didn’t come up with a 25-$30 [tasting] Central Coast Pinot,” he says.
Hughes sees Tastry as the “great de-risker.” Accurately identifying whether a wine will sell or not is the difference between success and failure for wineries, restaurants, and retailers. Consumers, meanwhile, often purchase wine not knowing if they will like it. Tastry can help with both.
“I look at [Tastry] as the ultimate matchmaker,” Hughes says. “It’s difficult to get consumers to buy. This allows you to know whether you have that customer and then for that customer to use the Tastry database to find that wine in your store.”
“A high degree of confidence”
Dai Deh, vice president of marketing and direct-to-consumer at Distinguished Vineyards in Novato, California, started using Tastry in 2023. Deh began incorporating Tastry insights into the product development process for a new wine brand. He used it to supplement the company’s more traditional process.
Dai Deh, VP marketing and DTC, Distinguished Vineyards
Traditionally, the company might create potential blends and ask employees to taste and score the wines. They would compile results and make decisions.
“We can send wines in to Tastry, they analyze it, and they spit back those same results with a high degree of confidence,” Deh says. “That takes out the subjectivity of it.”
Distinguished Vineyards has also used Tastry insights and results to help inform strategic blending decisions. The company’s winemaker created three potential blends. Tastry provided scores for those blends but also gave recommendations for ways to blend those three finished wines together to elevate how broadly appealing the wine would be.
“We ended up going with one of those recommendations,” Deh says.
The information has also informed Distinguished Vineyards launch strategy. “We wanted to have a strong pitch for a retailer buyer, helping them understand that we could bring in new consumers into the set by offering something different from what they already carry,” Deh says. “It gave us the confidence that consumers would love our wine, and we could make the case with buyers that their shoppers would repeat their purchase.”
“Not the evil AI”
Some might look at Tastry as a powerful new tool for the wine industry and for consumers. Others might see it as taking part of the art and skill of winemaking – or wine buying or reviewing – and assigning that responsibility to a computer algorithm.
“I think a lot of winemakers would probably look at this and go ‘Well, this makes me not needed,’” Hughes says. “That’s not true at all.”
Deh says the winemaking team at Distinguished Vineyards has been open minded. “I think they feel comfortable that this is about giving us more tools and more objective data that we can use to make decisions. It’s not replacing anything that they would be doing but just giving us a different lens into the consumer.”
Tastry’s Axelsson viewing analyses
Another concern is that focusing so heavily on consumer appeal might lead to regression to the mean. Wine is often most transcendent at its outer edges.
“‘More popular’ is not always better,’” Axelsson says. “There is a tradeoff between popularity and distinctiveness, and we focus on optimizing how popular versus distinctive a wine can be to ultimately be the most competitive in its category.”
However one looks at it, there is no question that Tastry will change the wine industry. Tastry already works with the majority of the largest wine producers as well as many smaller brands.
“I think people shouldn’t fear it. This is not the evil AI,” Hughes says. “You have to look at it as a tool and understand what it can do.”
Northwest Wine Report is wholly subscriber funded and is my primary source of income. Please subscribe to support continued independent content and reviews on this site.
To receive articles via email, click here.
No thanks.
Boy, I have never felt a connection with wine descriptions like “barnyard” or “black fruit” much less the “odd” descriptors. I have often wished for a less subjective description, even accounting for MY subjective and variable tastes. Maybe this is something that might help me. I don’t see it as cutting anyone out of the game, especially wine makers or reviewers. Hopefully, it can make what they do/say more reachable to us civilians trying to find a tasty treat to go with dinner
As a UCD educated sensory scientist, I bristle at the description of this firm calling itself a “sensory sciences company.” Rather, it seems to be a chemist using a computer to create big data and make predictions. This approach can narrow down choices but it can’t replace a trained human sensory panel or a well designed and executed consumer test.
Also the phrase “advanced GCMS-like technology” sounds like a black box. Garbage in, garbage out?
Many people feel they are qualified to “do” sensory just because they have a nose and a mouth. It’s a bit more involved than that. It’s called “sensory science” for many reasons.