AI assistant · N4 Executable plan
Automated Messaging Builder for Restaurants
Writes the messages your restaurant answers automatically -welcome, booking, delivery, opening hours, address, menu, events, frequent questions and follow-up- each with its exact trigger, its answer time and the border of what an automatic reply must never answer.
N4Executable plan
A plan with owners, deadlines and the metric you review. Not advice: an agenda someone can execute on Monday.
◆ This assistant 23 of the 164 published assistants are at this levelIs this for you?
What it does and what you get
- There are eight or ten questions your restaurant answers a hundred times a month and always the same way, answered by hand by whoever has the phone nearby, who is rarely the right person at the right moment.
- The expensive mistake is not failing to automate: it is automating too much and promising what cannot be delivered, because a message saying "of course, we'll see you tonight" without knowing whether a table exists turns a sale into an argument at the door.
- This assistant separates what does not change -hours, address, the menu, how to order- from what depends on today, writes the first part and gives the second an answer time, an owner and a handover to a person.
- The nine message types are one repertoire of the same method, not nine assistants: the trigger changes, the rule does not.
What you end up with
Five chained deliverables: the counted inventory of what people ask, the repertoire of the nine messages with trigger and answer time, the texts ready to paste, the handover to a person by time slot and the maintenance sheet with owner and review date.
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: "We answer the same thing all day on WhatsApp: opening hours, whether delivery reaches a given neighbourhood and whether there is a table. Two locations, and all I have is WhatsApp Business, nothing else to automate with."
Nobody was asked what people ask: the history was read. Between 21 July and 17 August, 709 messages came in through WhatsApp and Instagram across both locations, and they were grouped by question in the guest's own words.
| Question in the guest's words | Times in 4 weeks | Channel | Changes today? | Decision |
|---|---|---|---|---|
| "What time do you open?" | 168 | No | Automatic | |
| "Do you deliver to my neighbourhood?" | 131 | The area no, the shift yes | Mixed | |
| "Do you have a table for 6 at 8?" | 96 | Yes | To a person | |
| "Send me the menu" / "how much is the bandeja?" | 88 | Instagram and WhatsApp | No | Automatic |
| "Where are you? Is there parking?" | 74 | No | Automatic | |
| "Can you take 20 people on Saturday?" | 41 | Capacity no, the date yes | Mixed | |
| "Is there any sancocho left?" | 37 | Yes | To a person | |
| "Do you take bank transfers?" | 29 | No | Automatic | |
| 45 different questions, once each | 45 | Both | — | Out |
Eight questions carry 664 of the 709 messages, 94%. The other 45 are 45 different questions that showed up once: none gets automated, because automating what happens once is exactly the work nobody maintains and the one that six months from now will be saying something false.
Of those 664: 359 answer themselves in full without knowing anything about today (54%), 133 can never be answered alone —table and dish availability— and 172 are mixed, where the automatic reply hands over the fixed half and announces the missing one. Adding the fixed ones and the fixed half of the mixed, the first message settles or advances 531 of 664 conversations, 80%, without anyone on the team touching the phone. Illustrative figures, drawn from the history you provided.
SUPUESTO: there is no house figure for how many messages a restaurant this size receives per month, so the 709 threads counted over the 4 weeks of YOUR history are used, not an industry average. If the real volume were double, the midday answer time no longer holds with the hands available and somebody else has to be put behind the phone before the second message is switched on.
This is where you decide what goes out on its own and what does not, and there is one rule: if the answer changes with the day, no machine writes it.
| Question | What goes out alone | What the automatic reply does NOT say | Who answers the rest | Answer time promised |
|---|---|---|---|---|
| Delivery to a neighbourhood | The list of areas the in-house delivery covers, the minimum order and the usual delivery time | Whether a rider is free for that address at that hour today | Cashier of the location covering the area | 20 min outside lunch · after 2:30 p.m. during Tuesday-to-Friday lunch |
| Table for today | That groups of up to 8 are taken without booking and that the room fills from 12:30 to 1:30 | Whether a table exists at that hour | Floor manager of that location | 20 min outside lunch · 2:30 p.m. if they wrote between 12:00 and 2:30 |
| Whether a dish is left | That Bandeja de la casa, Sancocho de tres carnes and Trucha al ajillo run every day until they sell out | Whether any is left at this hour | Kitchen, through the floor manager | 20 min · after 8:30 p.m. nothing is promised: it is confirmed tomorrow at 11 |
| A party of 20 on Saturday | Capacity per location, the 28,000 to 72,000 COP per person range and that a deposit holds the date | Whether Saturday is free | Admin | Same day if they write before 5 p.m.; otherwise before 11 a.m. the next day |
The promise breaks exactly when most messages arrive. From 12:00 to 2:30 Tuesday to Friday and Sunday midday, the 18 operations people are serving and nobody looks at the phone: promising five minutes there is a lie the guest can time with a watch. So the automatic reply in that slot tells the truth —"we will get back to you after 2:30"— and the four o'clock one promises twenty minutes and keeps them. A long, honest wait buys more than a short, false one.
All of this fits in the app you already have: greeting message, quick replies with a shortcut and away message. Nothing needs to be bought.
Greeting message (sent on its own to anyone writing for the first time or after 14 days of silence):
Welcome to Sazón de Origen. Here we answer opening hours, address, menu and delivery straight away. If you need a table for today or want to know whether a dish is left, write "TABLE" and a person will take it: 20 minutes outside lunch, after 2:30 p.m. if we are mid-service.
Quick reply /hours:
We open Tuesday to Sunday. Lunch from 12:00 to 3:30 p.m. and evenings from 6:30 to 10:00 p.m. Sunday lunch only, 12:00 to 4:00 p.m. Closed Monday. Laureles and Envigado, same hours.
Quick reply /delivery:
In-house delivery to Laureles, Estadio, Floresta, Envigado centre and Sabaneta. Minimum order 45,000 COP, usual delivery 35 to 50 minutes. Outside those areas we arrive through the delivery platform. To confirm it reaches your address today, write "DELIVERY" and we will answer.
Quick reply /menu (names the three house dishes and not a single promotion, so the sales mix does not sink; the one that goes first is the one that leaves margin):
Here is the full menu. The house dishes: Sancocho de tres carnes, Bandeja de la casa and Trucha al ajillo. Prices run from 28,000 to 72,000 COP.
Quick reply /booking (asks for three details and not one more, and states the expected departure hour — which is what protects table turnover):
Glad to hold one for you. I need the date, the hour and how many covers. Tables are held for 90 minutes; if you plan to stay longer, tell me and we will look at it. A person confirms it, because availability changes every day.
Away message (10:15 p.m. to 11:00 a.m., and all day Monday):
We have closed. Tomorrow we open at 12:00 and answer from 11:00 a.m. If it is an order for tomorrow, leave it written here with the hour and we confirm it as soon as we open.
None of these six texts says whether there is a table, whether sancocho is left or whether delivery reaches you today: the four that depend on today go out through the keyword and a person picks them up.
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 pieces that hold the promise up. The handover table, slot by slot, with who is behind it, the time promised, what is not promised in that slot and what happens when the time runs out: it carries the slot where the honest answer is "after 2:30", the handover line that goes identical in all nine messages, the one-number rule with a label per location and the 10:15 p.m. cut-off with zero unanswered threads. And the maintenance sheet, with owner, last review, next date and the fact that expires each row early, plus the unknown-number test on the first Monday and the three figures read that day.
See the full example — free accountWorks with these AIs
How to use it
Which of your data it uses
This assistant works with 12 of the 23 fields in «My Restaurant», among them Restaurant name, Business type and City and country. 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.
v1.1.12026-08-26The body stops contradicting the answer ladder. It used to say «do not move on until it arrives», the very order the ladder was born to replace: the assistant asked the owner for what the owner does not have and stalled. It now says what the house doctrine says — ask one at a time, and if they do not know, continue with a marked assumption. FORMAT OF THE RESULT was not touched, so the example still corresponds and its seal moves up with the card.v1.1.02026-08-25Body rewritten to pay down duelo debt: it scored 78.5 out of 90.0, with `lexico` at ZERO out of 10. A card that decides what a restaurant answers on its own spoke of "questions" and "messages" without ever naming what is actually at stake behind each one: the dining room's occupancy, the booking's covers, table turnover, the delivery order ticket, whether the star dish is left, the contribution margin of the dish the menu message pushes, the sales mix that sinks if only the promotion is named, the channel's average ticket and the labor cost the fullest slot gives back. `cifras` was 3 out of 12 on a card whose promise IS an answer time: the 5 min promised against the 40 delivered, the 14-day indicator and the 30-day maintenance pass now come in. And `cuidados` (4 of 6) and `verificacion` (4 of 5) are completed with the nine-button-menu trap and five imperative checks. The declared N4 level is kept: the FORMAT still delivers chained blocks and not numbered pieces. Rises to 97.5.v1.0.02026-08-25The assistant is born.Same method, another variant
Inside the app this is a single control: you press it and the assistant reframes itself, including the variants that don't appear here because they don't have their own page yet.
More assistants in this category
Writes the replies to four and five-star reviews: sorts them by what each one brings, hands out a long reply o…
N4 · Executable planBuilds the flow your restaurant uses to win back a guest who left angry: who enters and who does not, the priv…
N4 · Executable planWrites the protocol for the first hours when something turns against you online: three levels with countable c…
Included in the always-growing library
Access to every published assistant, adapted to your AI and personalised with your restaurant's data, updates included.
