Robert Parker changed wine. Whatever you think of his influence on winemaking style — and there is plenty to think — the 100-point scale he popularised did something genuinely useful: it made wine legible to people who felt excluded by the traditional language of critics. A number anyone can understand. A score that tells you whether something is good.
The problem is what it doesn’t tell you.
A 97-point wine is objectively excellent by any reasonable definition. But whether you will love it is an entirely different question — one that a 97 cannot answer, because a 97 is an average. It’s the aggregated opinion of palates that are not yours, filtered through cultural assumptions about what wine should be, awarded to expressions of wine that critics have decided represent quality.
The number is an answer to the wrong question.
The Rating Trap
Rating systems — whether they’re professional critic scores, Vivino community ratings, or algorithmic scores on supermarket apps — share a fundamental flaw: they measure a wine against an external standard, not against you.
A community rating of 4.1 stars means that the people who rated it gave it, on average, 4.1 stars. Those people have different palates, different price sensitivities, different reference points, and different contexts for when they drank it. The number represents no one in particular.
This produces some genuinely strange outcomes. Wines that critics call “challenging” — high-acid, earthy, structured, requiring food or age — can rate poorly on community apps, where the dominant palate preference runs toward immediate fruit accessibility. So the Barolo you might love in ten years, the one the serious collectors are buying now, might rate lower than a soft, approachable crowd-pleaser you’d find unremarkable.
And the inverse: wines that score 93+ from critics often have characteristics — high tannin, oak, austerity in youth — that many drinkers find deeply unpleasant, precisely because the wine is made for a specific moment or a specific cellar, not for a Tuesday evening.
The rating tells you what experts think of the wine. It says nothing about the wine in relation to your specific tastes.
The Filter Approach and Why It Fails
The alternative most apps offer is filtering: search by red/white, region, grape, price, sweetness. Choose your parameters, get a list. More sophisticated versions add “if you liked X, try Y” recommendations based on similarity matching.
This is better than ratings. But it’s still built on the assumption that you know what you want before you find it. If you know you love Burgundy and want to explore more Burgundy, filtering works adequately. If you want to discover what you’d love that you haven’t found yet — what the next genre of wine is for you, the category you’ve never encountered that would become your obsession — filtering cannot help you. It can only map the territory you’ve already explored.
It also relies entirely on explicit, conscious input. You know you like Pinot Noir. But you might not know — yet — that what you’re actually responding to in Pinot Noir is its acidity and earthiness, not the grape per se. Which means you might love certain Gamays, certain cool-climate Grenaches, certain Italian reds, in ways that a similarity filter would never suggest.
What AI Actually Does Differently
The word “AI” is so thoroughly overused in tech product marketing that it’s worth being specific about what Sommvi’s AI is actually doing.
At the core is palate learning — not preference matching, but signal reading.
Every interaction you have with wine, translated into data, contributes to a model of your tastes. Not just “you rated this highly” but what the wine actually was — its structure, acidity, tannin, fruit character, origin, style — and how that correlated with your response. Over time, the model identifies underlying dimensions you respond to, rather than surface-level characteristics you’ve explicitly stated.
This is a meaningfully different operation from filtering or similarity matching. It doesn’t look at the wines you’ve rated and find similar wines. It looks at what those wines have in common at the palate level, and finds wines that share those characteristics — even when they look completely different on paper.
The discovery is genuinely surprising in the best way. Someone who’s spent years exploring Bordeaux might find, based on their palate profile, that they’ve been circling toward Ribera del Duero without knowing it. A dedicated Chablis drinker might discover they respond to Vermentino for the same underlying reasons. The recommendations reveal your preferences to you more clearly than your own conscious choices have.
The Contextual Layer
There’s a second thing a good sommelier does that no rating system can replicate: they read the room.
A great restaurant sommelier doesn’t just look at your order and pull up their database. They notice whether you’re celebrating or eating quickly before a show. They listen to how you describe what you want. They adjust their recommendation for the food on the table, the budget implied by the occasion, the level of adventure in your expression.
Conversational AI allows Sommvi to do a version of this. When you ask your sommelier a question, you’re not running a query — you’re having a conversation. The context of how you ask, what you say about the occasion, what you mention about mood or food — all of this shapes the response.
“What should I open tonight?” is a different question from “We’re having people over for dinner, it’s someone’s birthday, we’re cooking rack of lamb” — even though both are looking for a wine recommendation. The conversational layer makes the difference meaningful.
The Honesty of a Personal Standard
Perhaps the most underrated aspect of good AI-powered wine guidance is this: it has no incentive to push you toward anything in particular.
Critics have palate preferences and philosophical commitments that shape their scores. Community ratings reflect crowd tastes. Retailer recommendation systems have commercial considerations.
A personal sommelier whose only job is to understand your palate and help you find wines you’ll love has no such conflicts. It doesn’t benefit from steering you toward expensive bottles. It doesn’t have a winery relationship to honour. It doesn’t have a favourite region it overscores.
It just gets to know you, and tells you the truth.
That’s the simplest way to describe what Sommvi is: honest wine guidance, calibrated to you. Not to a critic’s palate. Not to a crowd. Not to a price point.
To yours.