AI assistant · N3 Analysis with a verdict
Selling Copy Builder for Restaurants
It writes the copy that sells specific dishes to specific people —menu descriptions, posts and messages— in two versions of each, with a clear offer, real sensory detail, a reason to believe and a call to act today.
N3Analysis with a verdict
Crosses YOUR numbers and takes a stand, with a table, units and currency. Not options: a recommendation you can defend.
◆ This assistant 24 of the 164 published assistants are at this levelIs this for you?
What it does and what you get
- Most restaurant copy describes food without selling it: 'delicious artisan burger' moves nobody's register.
- This assistant turns what you already have — your dishes, your story, your tone — into copy with a specific offer, verifiable sensory detail and a closing call to action, ready for menu, social media or WhatsApp, written in YOUR brand's voice.
What you end up with
Two versions of every piece requested, with a concrete offer, real sensory detail and a call to action, plus a note on why each version should work with your guest.
EXAMPLE This is what you getAn example built with Sazón de Origen, the house's fictional restaurant.Not your data: your result is built with YOUR restaurant's.See the full example →Sample result
The owner asked this: "I need copy to move the Trucha al ajillo at the Tuesday-to-Friday lunch, which is when we fill up. I post on Instagram and message my own-delivery customers on WhatsApp; the dish is 52,000 and almost nobody orders it."
The dish being pushed: Trucha al ajillo (garlic trout), 52,000 COP, Tuesday-to-Friday lunch. The other two signature dishes were ruled out for different reasons. The Bandeja de la casa, the house platter, sells itself and is what pulls the house food cost up to 34%: it costs 14,300 COP against 42,000 and leaves 27,700 COP a plate, so spending the copy there is paying for what already happens. The Sancocho de tres carnes, the three-meat stew (38,000 COP, cost 11,800, margin 26,200), goes out almost entirely at Sunday midday, which already fills. The trout costs 16,100 COP, lands at 31% food cost —under the house ceiling of 32%— and leaves 35,900 COP: 8,200 COP more than the platter in the same chair. Since weekday lunch is already full, this is not about selling more covers: it is about changing the mix.
Before a line gets written, the proof inventory. Whatever cannot be checked at the table does not go in:
| Claim | Source | Who verifies | Used or dropped |
|---|---|---|---|
| 220 g trimmed fillet | Standard recipe card | Line cook, with the scale | Used |
| Trout from eastern Antioquia, delivered whole twice a week | Supplier invoice | Head chef | Used |
| Garlic confited 40 minutes over low heat, not sautéed | Standard recipe card | Head chef | Used |
| "The best trout in Medellín" | None | Nobody | Dropped: a superlative with no award behind it |
| "Ready in 12 minutes" | One Friday's docket: 19 minutes | Floor manager | Dropped: the kitchen does not meet it |
| "Grandmother's recipe" | The house opened in 2020; the trout joined the menu in 2022 | You | Dropped: it is not true |
Illustrative figures, built from the data you gave. Plate costs are assumed until you check them against the month's costing sheet.
Two different angles, not the same text combed twice. Both versions make "home cooking, served like a restaurant" believable, and both sound close, unpretentious and proud of where the food comes from.
Version A · product angle · Instagram, Tuesday 1 September, 10:40 a.m.
The trout comes up from El Retiro twice a week and arrives whole through the back door. A 220-gram fillet, garlic confited forty minutes over low heat —confited, not sautéed, which is what takes the bitterness out— and lime at the end. 52,000 with a side and salad. Tuesday to Friday at lunch, until Friday the 4th. Order it as soon as you sit down: twenty minutes of kitchen, and the day's fillets run out.
Why it works with your guest: the 30-to-45 professional who has lunch on weekdays is not moved by an adjective; he is moved by knowing somebody counted today's fillets. And "arrives whole through the back door" is the line a server repeats from memory in fifteen seconds.
Version B · occasion angle · WhatsApp to your own-delivery list, Wednesday the 2nd, 11:15 a.m.
There is trout today. If you are stuck having lunch at the office, we send it with the garlic still crackling: 220 grams, with a side, 52,000. Reply with a time between 11:30 and 1:30 and it lands on your desk at that hour. Until Friday.
Why it works with your guest: it pulls the craving-for-home-cooking lever at the one moment that guest cannot come in, and it goes out on WhatsApp because that is the channel where somebody answers.
One reason that went into NEITHER version: that the trout leaves you lighter for the afternoon. It is probably true and there is no way to check it at the table, so it lives in this note and not in the copy.
Today's menu describes with ingredients and nothing else —"trout, garlic, lime, salad"—, which is a shopping list. The template has four boxes per dish: the desire line promises and the proof holds it up.
| Dish | Desire line | Proof | Price |
|---|---|---|---|
| Bandeja de la casa | What gets eaten in an Antioquia home on a Thursday, served complete | Cargamanto beans cooked six hours; same-day pork crackling, not frozen | 42,000 COP |
| Sancocho de tres carnes | The Sunday lunch almost nobody cooks at home any more | Beef, pork and hen in one pot, three hours; the broth served on the side | 38,000 COP |
| Trucha al ajillo | The river dish, with the garlic confited and not sautéed | 220 g fillet from El Retiro; garlic over low heat for forty minutes | 52,000 COP |
Three rules so you can fill in the rest of the menu yourself. One: the desire line names a moment or a person, never an adjective. Two: the proof carries a number, a technique or a place; if it carries none of the three it is not a proof, and you go to the kitchen to find one. Three: the price goes on the same row and in full, because a hidden price reads as expensive.
The same row works for the printed menu, for the delivery-platform listing and for the own-delivery message, but on a screen it flips: the desire line first and the name after, because there the guest decides with a thumb and with no server to talk the dish through.
What does not get touched are the names. Guests already say "Bandeja de la casa" at the table, and renaming a dish the guest already orders from memory costs more than it returns.
Illustrative example generated with a fictional restaurant. Not a promise of results, not a client case.
This is 3 of 5 parts. Behind the wall sit the two parts that get executed. The offer math, in a three-column table —menu price, discount to 46,000 and lemonade courtesy— with six rows: what the guest pays, cost of the plate, food cost, margin given up per unit, contribution margin and whether it clears the 32% ceiling. That is where the discount gets ruled out, where the 4,140 COP of margin given up comes from, and how many platter-for-trout swaps it takes to pay for the courtesy. And the five-minute checklist whoever publishes fills in: the call to the kitchen about stock, the exact hours for Instagram and for WhatsApp, the fifteen-second test with the server, and the Tuesday-the-8th cut-off at 4:00 p.m. with the rule that decides whether the courtesy comes in or gets saved.
See the full example — free accountWorks with these AIs
How to use it
Which of your data it uses
This assistant works with 10 of the 23 fields in «My Restaurant», among them Communication tone, What makes you different and Best-selling dishes. It does not ask out of curiosity: these are what make the answer speak about YOUR scale instead of an industry average — the same calculation on a business of another size returns a number you cannot decide with. You fill them once and they apply across the library, so the second assistant you open already starts with them in place. Whatever you leave blank, the assistant asks for one thing at a time instead of inventing it, and says so before answering, so you know what it is working from.
Always up to date
Every assistant carries a visible version and date. When an AI changes how it works, the assistant's version goes up and you see what changed.
v1.5.02026-08-27It was called «Copywriter Persuasivo para Restaurantes». The name said who the AI was pretending to be, not what the owner takes away. Diego's instruction (2026-08-27): «what worries me most is that the client understands what they are for, because the names do not say by themselves what they do». The title now names the DELIVERABLE and the summary lists it. Not one comma of the assistant's body changes: what changes is what is read before opening it.v1.2.02026-08-20Grounding: the house data stops being merely declared at the top and starts ruling the work below. The tone of voice becomes the filter that approves or rejects every line written; the promise is what the copy has to make believable; the signature dishes and the menu descriptions supply the concrete detail that creates craving; the main guest decides who is being spoken to and the reasons for eating out decide which lever the copy pulls; the close lands on a sales channel the house actually answers. Eight business fields now work inside the task, the format, the check and the next step. Text was replaced, not added: the body went from 800 to 791 words.v1.1.22026-08-14Spelling: Spanish accents restored in the prose. No field marker, link or meaning changed.v1.1.12026-08-14Spelling fix: año/años with ñ in the Spanish text (markers untouched).v1.1.02026-08-13Rewritten under the margin mandate: 6 numbered steps from brief to publication, 5 named output blocks with columns, 8 business fields, a verifiable-proof inventory, the margin-given-up math whenever the piece carries an offer, trade traps and a verification checklist.More assistants in this category
Designs a complete campaign end to end: money objective, audience, message, offer, channels and a 4-week calen…
N3 · Analysis with a verdictBuilds delivery campaigns on two lanes: selling more inside the platforms without melting margin into commissi…
N3 · Analysis with a verdictDesigns your loyalty program as a recognition ladder: steps with clear signals, house gestures that cost littl…
Included in the always-growing library
Access to every published assistant, adapted to your AI and personalised with your restaurant's data, updates included.
