Coverage before pairing. Every time, in that order.

The shift, in order

1. Five o’clockCoverage first. Two courses the bins cannot cover, one short for forty covers, and what to open instead. Before anybody prints a menu.
2. Pre-shiftA defended pour per course: the structural argument, the pairing it avoids and why, alternates by role, the tableside line, the pour size and the bottle count.
3. ServiceBin numbers, glassware, service temperature, and what to reach for if a table goes off-pairing. The pass view works on a phone in a dark cellar and works with no signal.
4. Close-outOne action. The bottles come off the record, and the valuation and the signals move with them. Nobody re-keys anything into a spreadsheet at one in the morning.

A pairing it will name as hard rather than solve

Artichoke, asparagus, raw vinegar, high heat, dark chocolate. Cynarin in artichoke makes the next sip taste sweet and hollow; asparagus does something similar with green, sulphurous notes. Aivin says the course is hard first, then offers the least bad option, rather than proposing a Grüner as though the problem were solved.

That is the difference worth paying for. Anything can produce a pairing. Knowing which courses do not have a good answer is what a sommelier does, and a tool that pretends otherwise gets overruled once and then ignored.

What it will not do

It will not pour a bottle the record does not hold, and it will not quietly substitute a non-compliant wine to avoid showing a gap. If a table’s constraints leave nothing on the list, it says so and names what to buy.

It will not overrule the somm. It argues, names the near-miss, and when you disagree it asks why and keeps your answer, attributed. Over a season those answers are the house’s judgment base — exportable, and yours.