The pairing library

Every pairing here is drafted by the press below and corrected at the bench; client builds are written against the client’s menu and stay theirs.

The dish

scored on ten axes

The engine

Proof #b0940e

drafts two tiers, deterministic

The bench

Christina tastes and corrects

The card

signed, laminated, on the pass

Set a dish · seven fixtures

Or move the axes

Dominant demand axis

umami

demand 0.63 = umami 0.90 × 0.70

Proof of pairing

Warm tomato consommé, kombu

amuse · japanese

Dish intensities: acidity 0.50 · richness 0.20 · salinity 0.35 · umami 0.90. Dominant demand axis: umami, demand 0.63.

Signature pour

Amuse Zero: Umami Accord

Verified 0.0%
kombu dashi (non-fermented, cold-brewed)60 ml
tomato water, clarified30 ml
still water30 ml

Batch of 8 at 120 ml · combine all components in a mixing glass over ice, stir 20 rotations, strain into chilled glass.

18°C / 64°F · celadon ceramic cup · toasted nori ribbon

Savory umami depth in the beverage amplifies the dish's umami, lengthening the finish.

Reserve pour

Amuse Reserve: Fermented Umami Study

<0.5% ABV class
NA sake-style (fermented rice water, <0.5% ABV)60 ml
kombu dashi (non-fermented, cold-brewed)30 ml

Batch of 8 at 90 ml · pour fermented base gently into chilled glass, add accent component, serve immediately without ice to preserve effervescence.

17°C / 63°F · celadon ceramic cup · toasted nori ribbon

Engine draft, unsigned

Proof #b0940e · Austin · 7-26

Seared foie torchon, verjus gel

Proof #69f04d · engine draft, unsigned

starter · french · test menu

Dish intensities: acidity 0.40 · sweetness 0.30 · bitterness 0.10 · richness 0.95 · salinity 0.50 · tannin 0.10 · spice 0.10 · herbaceous 0.10 · umami 0.60. Dominant demand axis: acidity, demand 1.48.

ink - the dish, 0–1 · red - demand on the glass, dominant = full measure · overprint - echo

Bright acidity in the beverage cuts the richness of the dish, resetting the palate between bites.

Signature pour

Starter Zero: Acidity Accord

Verified 0.0%
verjus blanc45 ml
yuzu citrus water (1:4 dilution)30 ml
still water45 ml

Batch of 8 at 120 ml · combine all components in a mixing glass over ice, stir 20 rotations, strain into chilled glass.

8°C / 46°F · Riedel Viognier white wine glass · lemon verbena leaf

Reserve pour

Starter Reserve: Fermented Acidity Study

<0.5% ABV class
NA sparkling wine (fermented, <0.5% ABV)60 ml
verjus blanc23 ml

Batch of 8 at 83 ml · pour fermented base gently into chilled glass, add accent component, serve immediately without ice to preserve effervescence.

7°C / 45°F · Riedel Viognier white wine glass · lemon verbena leaf

Smoked yuzu sorbet, shiso

Proof #2e7e13 · engine draft, unsigned

dessert · japanese · test menu

Dish intensities: acidity 0.90 · sweetness 0.60 · bitterness 0.20 · richness 0.20 · salinity 0.10 · tannin 0.10 · spice 0.10 · herbaceous 0.50 · umami 0.10 · effervescence 0.20. Dominant demand axis: acidity, demand 0.63.

Signature and reserve · identical to the draft above

A different dish plate, the same demand argmax: version one maps both test dishes to the acid accord. The red bar is what the correction pass exists to re-rank.

What a client library looks like

Four to eight courses, two tiers each: a signature pour, verified 0.0%, nothing in the glass ever fermented; and a reserve pour in the <0.5% class, one certified ferment for depth, labeled as what it is. Each build ships as a laminated spec card with batch quantities, prep timing, hold characteristics, glass, and garnish, plus the sourcing line and the cost line.

A ferment never gets labeled 0.0%, even when it measures 0.0 on paper; the classifier enforces the downgrade automatically.

Photographs of the bench - batches, temperatures, corrections in the margin - ship with the first paying room. Nothing here is staged in the meantime.

Start with the audit

Free, delivered before any call: a labeling review of what you pour now, three swaps you can make this week, and a pairing sketch for one or two signature courses. Request it on the menu page.