Menu Lab — the creator-distilled restaurant

The menu board

One creator per lane. A lane isn't "inspired by" — it's their full corpus distilled to taste-test winners, then kitchen-validated by us.

🥩

Steak program — Guga (Guga Foods + Sous Vide Everything)

788 videos distilled → the science + the recipe book. House picanha, sous vide service model, dry-age program, Butter of the Gods.

DISTILLED

🍗

Fried chicken — Joshua Weissman

556 videos distilled → the fried chicken program. Cured thighs, the craggy-flake dredge, Nashville hot-oil finish, double-fry wings, par-fry service model. He beat every fast-food chicken chain he rebuilt.

DISTILLED

🔬

Sourcing & technique science — Ethan Chlebowski

391 videos distilled → the value engine. Not a menu lane: a horizontal layer. Every "is expensive X worth it" verdict as a buying sheet, plus the food science that makes the other two lanes cheaper and more consistent.

DISTILLED

🍕

Pizza & the operator layer — Charlie Anderson

117 videos distilled → the operator. NY and Detroit canon, the bake-ahead-and-reheat service model, and the only real P&L in the project: he ran a pop-up for a year and opened Good Pizza in Cleveland. Also carries the cheesesteak sub-lane.

DISTILLED

🔥

Barbecue & the beef program — Mad Scientist BBQ

316 videos distilled → the control group. 85 single-variable tests. The 145°F hold protocol the whole project has been circling, the full yield chain (raw brisket is ~38% servable), and a cheap-beef program — clod at half brisket's price, chuck indistinguishable blind — that is entirely halal. Also the lane with no pork blocker.

DISTILLED

🍔

Smash burger — candidate: Weissman / George Motz

Guga's corpus already contributed: dry-aged-trim patties beat fresh, wagyu shoulder clod at 80/20, plain-patty compound butters. Lane owner TBD.

Chlebowski input: a three-tier ground beef test found no price-quality ladder — budget 73/27 browns and tastes great, lean is a calorie play not a flavour upgrade. His favourite patty was the half-smash, splitting the difference between crust and juiciness. Cheap beef, good technique.

OPEN

🍟

Fries & potatoes — candidate: Kenji López-Alt

Guga data points so far: one-by-one fried chips beat store-bought; duck-fat fries at the 9.1-rated steakhouse; beef-tallow steak fries. Kenji's potato science is the obvious corpus.

This lane is closer to done than it looks. Chlebowski ran Kenji's boil-flash-fry-freeze-refry method on camera across three separate videos and it held: salted and vinegared boil, ~60s flash fry at 375°F, freeze, finish ~4 min at 375-400°F. Russets only, and baked versions lost every time. A full Kenji run would add depth, but the method itself is already sourced and validated — including the freezer hold that makes it a service model.

OPEN

🍞

Bread — candidate: Weissman / Brian Lagerstrom

Milk bread already surfaced twice in the steak corpus (katsu sando, short-rib sandwich winner). A sandwich program needs a bread lane.

OPEN

🥣

Sauces & sides — partially seeded from Guga

Already banked: chimichurri (champion), mole, ssamjang service, sodium-citrate cheese sauce, garlic-confit spread, Mongolian sauce ("elevates any grade of beef").

OPEN

🍰

Dessert — TBD

Guga's own tour verdict is a hint: the thing he remembered from the Basque chef wasn't the steak, it was the cheesecake.

OPEN

Signature moves

Corpus-proven techniques we have not seen on a restaurant menu. These are the differentiators.

🥛 The surface-film sear

Whole-milk powder dusted pre-sear beat brown butter and control in Guga's test. Extra Maillard, no milky taste. Chlebowski reached the same place from six directions with mayonnaise — egg protein plus an oil that clings where oil alone runs off. The signature isn't milk powder, it's the film. Untested and obvious: both at once. The convergence →

🕰️ Dry-aged tallow baste

The "steakhouse scam" inverted into a feature: rendered dry-age fat basted on a fresh steak reads as a 35-day age. Dry-age flavor at fresh-steak food cost.

⚙️ The cook-ahead service model

Sous vide → ice bath → hold up to 6 days → reheat 135°F/30 min → sear at the pass. Tested equal-or-better than fresh. This is literally how a steak program scales to covers.

🧂 House umami salt

The homemade powder that beat MSG blind, cut 50/50 with salt (the winning seasoned salt). A signature finishing salt nobody can buy. Chlebowski supplies the mechanism: glutamate is amplified by the nucleotides in meat, mushroom, dried fish and aged cheese, which is exactly what that powder is made of. His own blend lands at 75:25 salt to MSG, dosed near 1% of weight.

🍚 The 48-hour "fast age"

Koji-pressed steaks that beat a real dry-age in a blind test. Dry-age menu line without 35 days of cabinet inventory.

🥚 Umami snow

Worcestershire-cured egg yolk grated tableside over the steak. Cheap, dramatic, corpus-validated.

🧊 The 145°F hold

Not a signature so much as the house architecture. A 12-16 hour hot hold measurably improves brisket beyond what the cook alone achieves, and Yoder proved it accidentally on paying customers when a batch held seven extra hours beat the lunch service. Cost: $350, or a 195°F oven for nothing. Protocol →

🌿 The post-bake finish

Anderson finishes every pizza after the bake (basil, pecorino, hot honey, chimichurri, stracciatella) and calls it a crucial part of what makes his pizza distinctive. It costs almost nothing and it is the one thing a held, reheated item cannot lose on the rack. Generalises to every cook-ahead lane we have.

The method

1 · Distill. A creator's entire catalog → caption transcripts → per-video verdict extraction (what won the taste test, exact numbers) → keep only winners. The pipeline is built and reusable; the Guga run took one evening.
2 · Book it. Winners become recipe cards with the verified temps/times and our proportions where the videos never stated amounts.
3 · Kitchen-validate. On-camera verdicts are not service-ready. Each dish gets real passes: does it hold, what's the food cost, can a line cook repeat it at volume. Nothing goes on the menu straight from a transcript.
4 · Credit loudly. The creators are the R&D lab and the story. The menu says so — it's the marketing, not a secret.

Reality notes

Where the thinking is honest with itself.

Adjacent things exist; the exact thing doesn't seem to. Creators have opened their own restaurants, and ghost-kitchen celebrity brands (MrBeast Burger) showed the failure mode: name first, food second. Aggregating many creators' validated technique with the food first is the differentiator. "No one's done it" is close to true but worth a real competitive scan before money moves. (Unverified beyond general knowledge.)
Validation path is cheap — and now it is priced. Charlie Anderson published his: free events for neighbours → paid pop-up → watch for repeat customers → only then raise money. The steak program plus one fried-chicken sandwich is already a pop-up menu.
The pop-up does not make money, and that is fine. Anderson's real numbers: ~$600 revenue a night on 20-25 pizzas, healthy 25.5% COGS, and still only $180 gross profit against 15 hours of work — $12/hour. Fixed costs and uncounted labour eat it. He treats the pop-up as paid market research, not a business. Anyone doing this for margin at that stage is misreading it. The full P&L →
Opening the actual restaurant: $100k+, and he still went ~$23k over. Eighteen months to find the space, three months to open it, ~200 applications for ~10 staff, 12-hour days seven days a week. The overruns were all hidden building systems (plumbing $7k, furnace $2.5k, cooler $3.5k). A second-generation space with a working oven and hood is the single biggest cost and time lever.
Hold-and-finish is now unanimous across all five lanes. Guga's sous-vide hold, Weissman's par-fry, Chlebowski's freeze-and-refry, Anderson's par-bake-and-reheat, Yoder's 12-16 hour hot hold at 145°F. Five creators, five cuisines, five independent arrivals. Nothing on this menu should be cooked to order from raw. The protocol →
Fire and labour are fixed costs, and that is the business case. Yoder: "it doesn't matter if I'm cooking one brisket or twenty, it's the same amount of time." Long-cook proteins have operating leverage that cook-to-order steak does not. Pair every slow expensive item with a fast cheap one (his chicken thighs at 62¢/lb) to absorb demand variance.
Cost the yield, not the sticker. Raw brisket is roughly 38% servable after trim, render and evaporation, so $3.49/lb is really about $9/lb on the plate. Every lane needs this done before any pricing conversation.
Consistency, not flavour, is the unsolved problem. Every lane here is a "what wins a taste test" answer. Anderson is the only source we have on making the 400th plate identical with someone else's hands, and his answer is unglamorous: fixed water temperature, timed mixes, written stretching guides, a caliper. Cross-lane, this is the biggest gap between our menu board and a real kitchen.
Sequencing: this lab develops the menu on nights and weekends. It stays behind boards (Aug 2026) and the Harmony close (Sep 1) in priority. The pipeline makes each new lane cheap to add.

Development log

2026-08-11 · run 5

Mad Scientist BBQ distilled: 316 videos, 85 single-variable tests → the control group. Five lanes live. This is the lane to build on. The cheapest cuts are the famous ones, the whole product is cooked ahead and held, and unlike pizza there is no pork blocker — he even cooked kosher brisket and said he "would not know the difference." Supplies the hold protocol (145°F, 12-16 hrs, $350 of gear or a 195°F oven for free), the full yield chain (raw brisket ≈38% servable), and a cheap-beef program: clod at half price, chuck indistinguishable from brisket blind, chicken thighs at 62¢/lb that beat brisket in his own customers' feedback. Hierarchy confirmed a third time: technique > equipment > grade.

2026-08-11 · run 4

Charlie Anderson distilled: 117 videos → the operator. Four lanes live. He is the first source in the project who actually opened the restaurant, and he published the numbers: the pop-up nets $12/hour, the buildout ran $100k plus ~$23k of unbudgeted plumbing and HVAC. Also: the reheat is a technique that changes the recipe, bake time beats ingredient spend, never buy a pizza stone, and fermentation peaks at 3 days. Kitchen test #3 is a blocker, not a curiosity — the Detroit par-topping method depends on pork pepperoni rendering fat into the crust, so a halal substitute has to be re-tested before this lane can ship.

2026-08-11 · tail batches

Folded the last Chlebowski batches into the value engine: nine more sourcing verdicts (salt is a scam above $1, onions and ground beef have no price ladder, never buy jarred garlic), sodium citrate as the real cheese-sauce answer, and the MSG spec that reconciles all three corpora at 75:25 salt to MSG. Two lanes moved: the fries method is already sourced and validated, and the smash burger has its beef answer. Kitchen test #2 identified: mayo + milk powder together as a sear film.

2026-08-11 · later

Chlebowski lane distilled: 391 videos → the value engine. Three lanes live. First real kitchen test identified: brown butter vs neutral oil basting, blind, on one cut — Chlebowski found no difference, Guga found brown butter wins.

2026-08-11

Weissman lane distilled: 556 videos → the fried chicken program. Two lanes now live. Cross-creator tension logged for kitchen validation: his steak test crowns tandoor + au poivre, Guga's crowns charcoal + picanha. Scouting list for run 3: Ethan Chlebowski (cross-cutting test science), Charlie Anderson (pizza/burger iteration quests), Mad Scientist BBQ (natively halal beef lane).

2026-08-10

Steak lane distilled: 788 Guga videos → science page + recipe book shipped. Menu Lab opened. Pipeline captured as a reusable playbook.