Meet the Woman Behind the AI that Tastes Wine

02 December, 2020
katerina-axelsson-tastry-wine-tasting-ai

Katerina started out analysing bag-in-box wine, a project she later shared with her college’s professor of technology Alex Dekhtyar. Enthused, he processed that data with artificial intelligence, and is now Tastry’s chief data scientist. By analysing the compounds in thousands of wines sourced from around the US from popular daily drinkers to the grandest of vintages, not only does Tastry help customers virtually try before they buy via the sensory science company’s Bottle Bird app, but its data can also guide wineries along their product development path, for example, reducing the quantity of Pinot Gris rosé in favour of upping the gallons of easy-drinking White Zinfandel according to customer purchases. 

And, with more than 11,000 wineries in the US alone, she adds that the Tastry lab has the capacity to test and add all their products to its database, with emphasis on supporting the little guys in the wine industry. “We’ve started working with smaller wineries that aren’t stocked by national chains so we can take their data and get smaller brands into consumers’ hands, which is the majority of the market.”

While sommeliers might get sniffy at the thought of a computer doing their job for them, Tastry means to complement the work of wine experts, rather than replace them. Katerina adds: “If someone set out to taste all the wines in the world, it would take them 18,000 years. So the somms we’ve talked to are very interested in the data in order to make intelligent business decisions. We’ve certainly tested more than any one person can taste, so it’s a valuable tool to better match customers.” 

New York-based sommelier Lauren McPhate, director of sales at Tribeca Wine Merchants in New York City, sees the positive side of improving service. She says: “It makes sense as an outside verification, like a score from a critic would. Anything that adds positive reinforcement to a standing recommendation is a positive in both of these fields – for somms on the floor and for wineries for sales, both direct to consumer, and to sales teams throughout distribution channels.” 

But Paz Levinson, cheffe sommelière exécutive for Groupe Pic based in Paris, is more cautious when it comes to pairings. She says: “My experience with ‘sensorial’ machines is that they can function in terms of recommending products but it is more delicate with dish associations. I have used them and the machine is more basic than a human.”

Tastry co-founder Charles Slocum believes technology can make a difference. He says: “We have a food-pairing engine so a restaurant can input and suggest wine pairings. We can help restaurants identify the palates walking in in an efficient way that makes people happy. Not all restaurants have sommeliers, and many servers don't have as much experience [as a sommelier], so it’s a useful tool for staff to provide better service.”

With patents in 54 other countries besides the US, the Tastry team is also hedging its sensory-science knowhow on other markets such as beer, food, perfume and cannabis. Katerina says: “Our next focus is definitely beer because it’s an industry we can understand through wine. Beer is a larger industry and also more fragmented, as there are very large and very small breweries, plus it’s localised thanks to the rise in micro breweries. Add in larger [bottle] formats and premium products – the selection is enormous.”

The vast world of food, recipes and ingredients has also started to form part of Tastry’s extensive flavour database. The sensory-science company recently worked with a packaged foods company keen to identify key flavours in sauces. Katerina says: “We analysed 128 sauces for a chicken wing company interested in targeting their market to consumers who would like one sauce the most, thereby eliminating having too many options.”

While the human connection between diners and staff at a restaurant is unlikely to be replaced by a robot (anyone remember Cynthia’s CyberBar under London Bridge? The robot Cyn has long been in retirement), Tastry's application might offer consumers a technological hand so they can better navigate their way around the often-overwhelming world of wine. As Katerina says: “We hope this technology will provide a way for people to discover wines.”

London Growing Avocadoes by 2050

Londoners Could be Growing Avocados by 2050

Next Article