AI for restaurants: what it can do for you today
By Alex Vilches, founder and CEO of Inmersa360 · · 7 min read
AI for restaurants can already serve every guest in their own language, recommend dishes based on what they fancy and their allergies, suggest the wine or the dessert and get the order ready in Spanish for the waiter. It works as a virtual assistant on the guest's phone, opened from the table with nothing to download, and leaves your team free for whatever needs a person.

Talk of artificial intelligence in hospitality mixes up very different things: kitchen robots, receptionists that answer the phone, software that forecasts stock and assistants that serve guests at the table. This guide sticks to what a restaurant, bar or café can put in place today for its guests, and what's worth preparing first.
Why now
Spain welcomed 96.8 million international tourists in 2025, a record according to the INE, Spain's national statistics office, and they spent more than €134.7 billion, according to the Spanish government. Many of them sit down on your terrace without understanding the menu.
On the other side there are more than 300,000 hospitality businesses, according to the Hostelería de España 2025 yearbook, competing for those guests with teams that can't keep up in high season. Guest-facing AI helps exactly there: it explains, in each guest's language, what your staff can't explain at every table.
What AI can do in a restaurant today
- Serve each guest in their language. It understands what a German or Swedish guest asks and answers in their language, by text or voice.
- Translate the menu by understanding each dish. Not word for word: it explains what an arroz del senyoret or a marinera is, the way a native speaker would.
- Recommend based on what they fancy. Meat, fish, something to share or surprise me, with the best matches shown first.
- Filter allergens. Guests mark what they can't eat and the menu hides those dishes, always with a reminder to tell the staff about a serious allergy.
- Suggest the wine or the dessert at the right moment, when the guest adds a dish, which is when the average spend is easiest to raise.
- Get the order ready for the waiter in Spanish, even if the guest chose it in their own language.
- Ask for a Google review at the end, with a button that opens the venue's profile directly.
- Tell you what happens at your tables: what gets ordered most, from which languages and what people ask.
Away from the table there are other uses, such as taking bookings by phone, replying on WhatsApp or forecasting stock. They're different tools, each with its own provider. Start with the problem that costs you most today.
The four most common ways to use it with guests
| Tool | Where it works | What it's for | Worth knowing |
|---|---|---|---|
| QR menu with automatic translation | The menu, on the phone | Reading the menu in another language | If it translates word for word, guests still don't know what the dish is, and it can't answer questions |
| WhatsApp chatbot | The venue's WhatsApp | Bookings, opening hours and questions before the visit | Built for before the visit, not for when the guest is already seated |
| AI phone receptionist | The phone | Taking bookings and answering calls | Useful if you miss calls mid-service; no help at the table |
| Virtual assistant at the table, such as ChefBot | The guest's phone, from the table stand | Recommending, answering menu and allergen questions and preparing the order | Needs your menu and allergens checked when it's set up |
What AI shouldn't do on its own
- Decide on a serious allergy. It can filter the menu with your information, but a guest with a serious allergy needs to talk to your team.
- Make things up. If something isn't on your menu, the assistant has to say so, not guess.
- Take payment at the table without control. All it takes is someone moving a table stand for orders and payments to get mixed up. Leave payment to your team.
- Replace the personal touch. It takes the repetitive questions. The welcome and the house recommendation are still your waiter's.
How to get it running in your venue
- Your menu up to date, with dishes, prices, photos and each dish's allergens. The list is in the 14 allergens that must be declared.
- What guests ask most. The wifi, whether something is spicy, what's in a dish or which wine goes with the meat. That's the first thing it has to answer well.
- Access from the table. A stand with NFC and QR, so guests open it by tapping their phone or scanning, with nothing to download.
- Tone and limits. Formal or friendly, and what always goes to a person.
- A week of testing with your team before it goes on every table. Ask it what guests ask you every day and fix whatever goes wrong.
With ChefBot, we do that work for you: we have it running in your venue within 15 days, with your menu, your branding and your allergens checked with you.
What the law says
Since 2 August 2026, the EU Artificial Intelligence Act requires people talking to an AI to be told it's an AI, unless that's obvious. Under Article 50, the obligation falls on the provider, but it's worth checking that your assistant introduces itself for what it is.
Allergens follow the same rules as on paper: Regulation (EU) 1169/2011 requires you to declare them, and AI is one more way of doing it, not an exception. It's explained in allergen menu rules. If the assistant handles personal data, the General Data Protection Regulation applies as well.
What to ask your provider
- How it introduces itself to guests and whether it makes clear it's an AI.
- Where its answers come from and what it does when it doesn't know something.
- Who checks the allergens and how long a menu change takes to show.
- What data it keeps and whether you can see statistics for your tables.
How to tell if it's working
Look at how many tables open it, in which languages, which dishes get added and what people ask. Compare average spend and Google reviews before and after. At Luna Creciente, in Torrevieja, Google reviews doubled in a few months thanks to the review button on its digital storefront.
The questions also show you what's missing from your menu: if lots of guests ask what a dish is, its name isn't clear.
ChefBot, Inmersa360's AI for restaurants
ChefBot is the AI virtual assistant that Inmersa360 builds in Torrevieja (Alicante, Spain) for restaurants, bars and cafés. It lives in the venue's digital storefront and in its AI menu, and guests open it by tapping their phone on the table stand or scanning the QR code. It already serves guests at venues such as Porthus, Las Cañas, El Charro Negro, Luna Creciente and Kafeina, which you can see in our case studies.
If you run a hotel, the same idea for your guests is AlojaBot. We explain it in AI in hotels.
We'll show you ChefBot running with your restaurant's menu, in your guests' languages.
In short
Start with what guests ask most and with your menu and its allergens. Choose easy access from the table, with no downloads. Decide what always goes to a person. Check that the assistant says it's an AI, and track languages, average spend and reviews.


