Wine & Spirits - Fall 2017 - 19

Wine Ring. This company has been around since
2011, raised over $4 million, and has hired a bouquet of Master Sommeliers and PhDs. The team has
tasted and catalogued more than 100,000 wines on
40,000 taste attributes. The app gathers information on wines you like when you snap a picture of
the label (optical recognition takes it from there) or
input the wine name manually. Users can then rate
the wine on a four-point scale ranging from "love it"
to "don't like it."
While the app does have a large database of wines,
it is not comprehensive. But there is the chance to
add a wine and rate it so that it can still be in your
personal history-keeping a log is always helpful.
However, after I'd rated 25 reds, whites and rosés,
the app only recommended red wines to me from
appellations I had input. That seems cautious and
facile for a team with so much brainpower. Further,
the app seems underdeveloped-there's no social
media sharing (though you can connect with fellow
users) and the five restaurants in the app have content that is out of date.

Vivino. The company boasts a large base of over five
million users in the US alone. The app is an easy way
to keep a log of bottles: Once the data is input manually or through an optical scan of the label, users
rate wines on a five-point scale (with half-point increments). It encourages registration, finding friends
and then seeing their recently liked bottles in a feed.
Throughout, the app suggests retailers where those
bottles can be purchased; the suggested retailer is
often the closest but not the cheapest.
On recommendations, the app is weak. After I had
input 25 reds, whites and rosés, the suggested wines
yielded not a single one that I liked (or even thought
I would like); it also offered several wines that I have
had and disliked. For pages and pages. Unlike Wine
Ring, which seemed to key in regions or styles that
I had rated highly and then offered somewhat simplistic lateral recommendations, Vivino actually suggested wines in styles that I gave one star on the app.
If this is the best AI that money can buy, they're going
to need more money.

Bright Cellars and Winc. These two websites offer
the promise of monthly "curated choices" (Winc,
formerly Club W) or "wines that you'll love" (Bright
Cellars). Each site has users fill out a brief questionnaire about food likes and dislikes and then generates

four wine recommendations. The picks feel canned.
Bright Cellars suggested two red blends, an Aussie
shiraz, and a California sauvignon blanc to me. The
chances that I would select four such wines in a wine
store are precisely zero. And how they came up with
these based on whether I like dark chocolate, orange
juice and strong tea seems like a reach. So I ran the
test again using different responses to the questionnaire and two of the four wines the engine generated
were the same.
Similarly, Winc's recommendations (two whites,
two reds) also seem pre-selected and only tangentially related to whether I take my coffee black or
love raw berries or mushrooms, let alone my actual
wine preferences. The company founders set up their
"direct-distribution wine company" (read: you can't
find these wines elsewhere) with the aim to make
wine "simpler to get and easier to enjoy." Both companies say they incorporate user feedback on wines
sent whn choosing wines for future mailings. Both
companies push people to buy with countdown
clocks during the order process and follow-up emails.

Another type of service uses a DNA sample to test for
certain hard-wired taste proclivities. Unfortunately,
my double helix couldn't be sequenced because the
turnaround time exceeded my deadline. This may be
an issue, too, for people not on a deadline: it takes
four to six weeks and costs $99 (and includes a bottle
of wine). I do wonder how revealing it would be; one
significant issue with this is that DNA is fixed, while
taste preferences may change over time. For starters,
context matters: I like to contemplate a concentrated
St-Joseph in the winter but am fine with a carefree
rosé in the summer. Also, many consumers get into
wine with simple "gateway" wines only to graduate
to something else and perhaps something else again
as palate and preferences (and pocketbooks) evolve
over a lifetime.
Judging from this selection, the robots are not coming for wine jobs yet-sommeliers and restaurant
staff can breathe easy for now. If a vexed consumer
were to give a skilled wine professional the name of
two or three wines they like, the recommendations
would start flowing faster than wine from a magnum. These electronic services may be useful for
other things, but when it comes to great, personalized wine recommendations-the Holy Grail-the
quest continues. ■

WINE & SPIRITS

FA L L 2 0 17

Table of Contents for the Digital Edition of Wine & Spirits - Fall 2017