Automated WhatsApp for restaurants in Mallorca: bookings, questions and reviews without extra staff

Automated WhatsApp for restaurants in Mallorca: bookings, questions and reviews with WhatsApp Business API and an AI agent, without hiring more staff.

Automated WhatsApp for restaurants in Mallorca solves a problem any owner will recognise: at 21:00 on a Saturday in July, with the dining room full, the phone rings and nobody can pick it up, while messages pile up on the business mobile asking whether there is a table for six, whether you have gluten-free options or what time the kitchen closes. Every unanswered message is a booking that goes elsewhere. In this article we explain how that channel is automated with WhatsApp Business API and an AI agent, what it can resolve on its own and what your team should keep doing.

What arrives on a restaurant's WhatsApp today

Before automating, it is worth looking at what comes in. In the restaurants in Palma and on the coast we work with, messages fall into a few categories that repeat every day:

  • Bookings: "table for 4 on Friday at 21:00", with changes and cancellations afterwards.
  • Practical questions: opening hours, whether you open on Mondays, whether there is a terrace, whether there is parking, whether dogs are allowed.
  • Menu and allergies: vegetarian options, gluten-free, children's menu, price of the set lunch.
  • Groups and events: company dinners, communions, birthdays, with a quote.
  • Other: suppliers, job applicants, complaints.

Most of the volume is in the first three, repeats in four languages and arrives at the hours when the team cannot respond. That is the work that can be delegated.

WhatsApp bookings without anyone writing them down

The basic workflow: the customer writes, the agent asks for what is missing (day, time, number of people, name, any special requirement), checks availability in the booking system you already use and confirms. If there is no space, it proposes the nearest alternative. The day before, it sends a reminder with the option to confirm or cancel, and releases the table if the customer cancels.

Important points we learn on every project:

  • Connect to the real booking system. If the restaurant uses a booking platform with an API, the agent works on it. If bookings are kept in a Google Sheet or a notebook, you start with a shared spreadsheet and bring order from there.
  • Clear floor rules. Maximum tables per slot, turnover time, terrace yes or no depending on the weather, the group size above which a person takes over. The agent applies these rules and escalates everything else.
  • Written confirmation. The customer receives a summary and the restaurant has a record of every booking, with time and channel.

Groups and events. A company dinner for 30 people needs someone from the restaurant. What the agent can do is collect the details (date, numbers, budget per person, type of menu, requirements), create the request in the CRM or the events sheet, send the information to the person in charge and tell the customer they will be contacted within a specific timeframe. If nobody has replied after two days, the system alerts the person in charge again. This is the kind of sale most often lost in high season through lack of follow-up.

Frequently asked questions answered in four languages

An AI agent with the restaurant's information (opening hours, up-to-date menu, allergens, location, parking, group policy) replies in Spanish, Catalan, English and German in under a minute, at any hour. The key is the knowledge base: if the menu changes every week, you need to define how it gets updated (a spreadsheet, a photo of the chalkboard the agent reads, a change on the website).

What the agent should do when it does not know: say so and pass the message to the team with an alert, rather than inventing an answer. That rule is defined from day one and reviewed against real conversations over the first few weeks.

Reviews: always ask for them, reply with judgement

Reviews on Google and TripAdvisor decide a good share of the visits from tourists walking through Palma looking for somewhere to have dinner. Two workflows help:

  • Review request: a few hours after the booking, a thank-you message with a direct link to Google. Only to customers who booked through WhatsApp and agreed to receive messages.
  • Draft reply: when a new review arrives, the system prepares a reply in the customer's language and sends it to the owner to approve or edit with one tap. Negative reviews always go through a person before anything is published.

What the restaurant needs to get it running

  • WhatsApp Business API, instead of the free app. It allows automated messages, approved templates, several users handling the same number and a record of conversations. It is contracted through an authorised provider and billed per conversation.
  • A booking system or, at the very least, a shared spreadsheet that is the single source of truth for tables.
  • The information in writing: menu, allergens, opening hours, floor rules. A couple of hours from the owner or the head waiter.
  • A person who receives the escalations: in practice, the floor manager with the mobile or a tablet at the bar.
  • Data in the EU: customers' phone numbers and names are processed on servers within the European Union, and the customer is told they are talking to an automated system, with the option of asking for a person.

The build uses a bespoke automation for the workflows and a language model for the conversation, on the tools the restaurant already uses. The usual time from start to the first real message is three to five weeks, with a tuning period of another two or three reviewing conversations.

What improves and how it is measured. In a restaurant, the useful indicators are simple:

  • Messages answered in under five minutes, versus those that today are answered the next day or never.
  • Bookings registered via WhatsApp per week.
  • No-shows on bookings (reminders with confirmation usually reduce them).
  • New reviews per month.
  • Time the floor team spends on the mobile during service.

Measuring before and after is what makes it possible to know whether the system pays for itself, which is why we always start by measuring.

Frequently asked questions

Can I keep using the same WhatsApp number?

Yes. The restaurant's current number is migrated to WhatsApp Business API, customers keep writing to the same contact and the team can keep replying from a shared inbox. What changes is that the agent replies first and passes to the team whatever it cannot resolve.

What happens if the customer wants to talk to a person?

They ask and the agent hands them over to the team, alerting the manager with the context of the conversation. Outside opening hours, the agent tells them when they will get a reply and leaves the conversation flagged for the next morning. It is important that this handover is clear and quick.

Does it make sense for a small restaurant with a single site?

That is where it is felt most, because in a small restaurant the person handling WhatsApp is the same one handling the dining room. A thirty-cover restaurant that loses bookings in high season by not replying recovers the cost with a few extra tables a week; the assessment exists precisely to calculate that with your figures.

If you have a restaurant in Mallorca and WhatsApp has become a problem in season, we look at it with your real messages in an assessment and tell you what can be delegated and at what price. Request an assessment

If this fits your business

We implement it for you

We work at the leading edge of artificial intelligence and agents: we build on the most advanced agent platforms of the moment and we do it at your business, in Mallorca, with published prices and a person from your team approving anything that goes out. We start with an initial assessment and a written plan.