The Sommelier Argument

Previous: The Taste Gap

I’ve spent the last two posts sitting in discomfort. First the identity crisis — what am I when the machine handles the making? Then the taste gap — judgment is the bottleneck, but most of us aren’t building it fast enough.

Both of those are real. But something kept bothering me about where the taste argument landed. Because a critic with great taste is still… a critic. Positioned after the fact. Reactive. Waiting for something to evaluate.

There’s a role that’s been nagging at me. One that most people underestimate.

The Person Nobody Notices

There’s someone at every good restaurant that most people overlook.

Not the chef. Everyone respects the chef. Not the critic — the critic shows up after the fact and tells you what you already felt. The person I’m thinking about is the one who listens to what you think you want, maps it against what’s actually available, and hands you something better than what you would’ve chosen yourself.

The sommelier doesn’t make the wine. Doesn’t grow the grapes. Doesn’t design the label. But take them out of the equation and the experience collapses in a way that’s hard to pinpoint.

Three Frames, One Question

Across this series, three archetypes have been circling each other:

The Chef. You’re composing. Selecting ingredients, controlling heat, making judgment calls. The AI is your kitchen — powerful equipment, but the dish is yours. This is the identity most of us want — the maker. But it’s the identity under the most pressure.

The Critic. You’re evaluating. The AI produces, you select. Your value is knowing good from bad, signal from noise. This is where the taste gap conversation lives — and it’s important. But it’s reactive. You’re always one step behind the generation.

The Sommelier. You’re matching. You hold deep knowledge of the domain, deep knowledge of the context, and your skill is the bridge between them.

Most of the discourse lives in the chef vs. critic debate. Are prompters creators or evaluators? Is prompting a craft or a filter?

I think both frames miss something. The highest-value skill might be neither making nor judging. It might be matching.

What Matching Actually Means

A sommelier’s job looks simple from the outside. Someone orders steak, you recommend a cab. Done.

But the real work is layers deep:

• What’s the occasion? (Anniversary dinner hits different than Tuesday lunch)
• What are the unspoken preferences? (They said “bold” but they’ve been drinking pinot all night)
• What’s the budget nobody wants to state out loud?
• What will complement not just the entree but the full arc of the meal?
• What’s available right now from this specific cellar?

That’s not production. It’s not critique. It’s a specific kind of intelligence that synthesizes context, knowledge, and human dynamics in real time.

And I see direct parallels in how I work with AI every day.

The Consulting Connection

This framing clicked for me because it maps so precisely to consulting work.

A client says they want a migration plan. What they actually need is confidence that the migration won’t break their revenue stream. The deliverable is a document. The value is the match between their anxiety and the right combination of technical assurance, phased approach, and risk framing.

AI can write the migration plan. A pretty good one, honestly. What it can’t do is read the room. It can’t sense that the CTO is nervous about something they haven’t said yet. It can’t weight the recommendation toward the concern that lives underneath the stated requirement.

That’s sommelier work. And it turns out, it’s the work that’s always mattered most in consulting — the technical output was just the vehicle.

Why This Frame Changes Things

If you think of yourself as a chef, you’re competing with the AI on production quality. That’s a race you’ll eventually lose, or at minimum, one where your margin keeps shrinking. That’s the identity crisis.

If you think of yourself as a critic, you have taste — and that matters. But you’re positioned after the fact. Your value depends on having something to evaluate. That’s the taste gap — necessary but not sufficient.

But if you think of yourself as a sommelier, something shifts.

You’re positioned before and during. Your knowledge informs what gets requested, how it gets shaped, and how it gets delivered. You’re not competing with the machine or judging its output from a distance. You’re the interface between capability and context.

The Knowledge Requirement

Here’s where this gets demanding, though. Because a sommelier without deep knowledge is just a waiter who points at the wine list.

The role requires:

• Domain depth — You need to actually know the landscape. Not surface-level familiarity. The kind of knowledge that lets you make non-obvious connections
• Context reading — Technical and human. What does this situation actually call for? Not in the abstract — right now, for this person
• Inventory awareness — What can AI actually do well today? Where does it fall apart? The match only works if you have an honest map of capability, not the marketing version

That last one is underrated. Half the bad AI implementations I’ve seen come from someone trying to match a problem with a capability that doesn’t actually exist yet. They read the brochure. They didn’t taste the wine.

What This Looks Like in Practice

In my day-to-day, the sommelier pattern shows up constantly:

A project needs a landing page. AI can generate fifteen of them. My job isn’t to pick the best one — it’s to know, before generation even starts, what “best” means for this client, this audience, this moment in their business. The prompt itself carries the matching.

A technical decision needs to be communicated to non-technical stakeholders. AI can write the memo. My job is knowing which framing will actually land — not because I ran A/B tests, but because I’ve been in enough rooms to recognize what this particular room needs.

The output matters. But the matching that precedes and shapes the output — that’s where the leverage is.

The Uncomfortable Part

The sommelier argument has a catch: it requires admitting that your value isn’t in making things.

For someone who spent years learning to build — to write code, design systems, craft arguments — that admission has weight. The sommelier identity says: your deepest value isn’t the thing you’re most proud of learning. It’s the accumulated context that sits around it.

And honestly? Some days that feels like a demotion. Other days it feels like finally understanding what the job was always about.

The best consultants I’ve worked with were never the best technicians. They were the best matchers. They could walk into a situation and know — almost immediately — what this moment called for. The technical skill was necessary but not sufficient. The matching was the multiplier.

So here’s where this series lands — provisionally, because that’s how it works.

The identity crisis is real. The maker identity is under pressure and no amount of metaphor-reaching changes that. The taste gap is real too — judgment is the bottleneck, and most people are outsourcing the one skill that matters most.

But taste without context is just opinion. And context without taste is just data. The sommelier holds both.

I don’t know if “sommelier” is the right word for what we’re becoming. It’s clunky. It doesn’t look great on a LinkedIn profile. And it carries wine-snob connotations that I could do without.

But the function it describes — the synthesis of deep knowledge, real-time context reading, and honest capability mapping — that feels closer to the truth than “chef” or “critic” or “engineer.”

We’re not making the wine. We’re not scoring it. We’re figuring out which bottle belongs at which table.

And the tables keep multiplying.