In 2025 the average no-show rate in Spanish restaurants fell to 3.3% (down from 3.6% in 2024), according to TheFork reservation data reported by InfoHoreca and Vinetur in March 2026; even so, those empty tables cost more than €15,500 a year in lost revenue per restaurant, according to El Economista (July 2026). The same data show that an automatic reconfirmation before the booking brings that rate down to 1.52%, and a card guarantee down to 0.66%. Spain also closed 2025 with a record 96.8 million international tourists (INE, Frontur, provisional figure from January 2026), which keeps the share of guests booking and asking questions in a language other than Spanish high. Summum IA implements conversational agents — by voice and text — that confirm and remind guests of bookings to curb no-shows, serve international guests in several languages, and anticipate demand by service and shift, without replacing the POS or PMS the establishment already uses.
A restaurant's phone or a hotel's switchboard receive the same recurring pattern of calls: confirm a reservation, ask about the menu or availability, change the time, cancel at the last minute without notice. Every one of those unanswered calls outside business hours is, in practice, a booking that goes to the establishment next door. An AI agent that calls or answers, confirms the reservation and reschedules it if needed, attacks the root cause of the no-show directly: according to TheFork's 2025 data, an automatic reconfirmation before the appointment reduces the baseline 3.3% absence rate down to 1.52%, and a card guarantee brings it down to 0.66%. In a restaurant handling 500 bookings a month, the difference between those rates translates into several recovered tables every week without anyone on the team making a single reminder call.
Summum IA implements the conversational agent integrated with the establishment's booking system — a hotel PMS (Mews, Cloudbeds, Opera Cloud) or a restaurant reservation engine (TheFork, Cover Manager, an in-house system) — and with the floor POS or ERP. The same agent serves guests in several languages: Spain received a record 96.8 million international tourists in 2025 according to INE, with the UK, France and Germany remaining the three main source markets, which means a significant share of calls and messages arrive in English, French or German. Beyond confirming bookings, the agent can recommend dishes based on allergies, preferences or kitchen availability, and feed a demand forecasting module by service and shift — breakfast, lunch, dinner, weekend versus weekday — to plan purchasing and staffing ahead of time, without duplicating the generic forecasting engine Summum IA already offers for other sectors.
Article 50 of Regulation (EU) 2024/1689 (AI Act), fully applicable from 2 August 2026 under its Article 113, requires clearly informing anyone interacting by voice or chat with an AI system when this is not otherwise obvious; the call-start and conversation-start notice ships configured from day one. Processing the guest's data — name, phone number, dietary preferences, booking history — is also subject to the GDPR (Regulation EU 2016/679): defining legal bases, retention periods and the data-processing agreement with the technology provider is covered by Summum Consultoría's sister page specialised in hospitality, while this page focuses on the conversational-agent and demand-forecasting layer.
We analyse call and booking volume by channel, the establishment's current no-show rate, the most frequent guest languages, and the shifts with the greatest demand variability, to prioritise the agent's design.
We define the booking confirmation and reminder blueprint, the escalation conditions to front desk or floor staff, and the languages the agent must cover based on the establishment's actual guest profile.
We connect the agent to the hotel PMS or the restaurant reservation engine (Mews, Cloudbeds, Opera Cloud, TheFork, Cover Manager or another system with an API) and to the floor POS or ERP, so it confirms, changes and cancels bookings in the same system the team already uses.
We configure the dish recommendation module based on allergies and kitchen availability, and the demand forecasting module by service and shift, fed with the establishment's own booking history.
The operational detail: what we deliver as part of the work and what we keep alive afterwards.
Confirmation and reminder blueprint
Documented diagram of the booking reconfirmation flow, escalation conditions to front desk or floor staff, and guarantee rules (card, prepayment) where the establishment requires them.
Multilingual conversational agent
Voice and chat system trained on the establishment's menu, schedule and processes, operating in the languages most in demand among international guests.
PMS/booking engine and POS integration
Connector to the existing hotel or restaurant booking system and to the floor POS/ERP, with no need to migrate platforms.
Menu recommendation module
Dish recommendations based on allergies, preferences and real kitchen availability, built into the same booking or enquiry conversation.
Demand forecasting by service and shift
Demand estimate for breakfast, lunch and dinner, and weekend versus weekday, based on the establishment's own booking history.
Documented AI Act compliance
AI-interaction notice under Article 50 of Regulation (EU) 2024/1689 from the first call or message, with an audit log of the conversation.
The AI agent for hospitality builds on Summum Sistemas' hospitality ERP/POS for day-to-day operations and on Summum Consultoría's guest data protection, closing the loop between the conversational booking, floor management and regulatory compliance.
Summum Sistemas implements the hospitality ERP and POS — floor, kitchen, stock and billing control — while this page covers the conversational-agent layer that confirms bookings, recommends menu items and forecasts demand: split by layer, no keyword overlap.
View service → consultoríaSummum Consultoría covers GDPR compliance for guest data — consents, retention periods, the data-processing agreement with the agent's provider — while this page focuses on the conversational automation of bookings and guest care.
View service →According to TheFork's 2025 booking data reported by InfoHoreca and Vinetur in March 2026, Spain's average no-show rate was 3.3% (down from 3.6% in 2024); an automatic reconfirmation before the booking brings it down to 1.52%, and a card or prepayment guarantee down to 0.66%. Summum IA's agent applies exactly that conversational reconfirmation mechanism in the window before the reservation.
It's configured to match the establishment's actual guest profile. Spain received a record 96.8 million international tourists in 2025 — according to INE (Frontur, provisional figure from January 2026) — with the UK, France and Germany as the main source markets; the agent is typically trained in Spanish, English, French and German, and extended to other languages based on the establishment's specific demand.
Yes. Article 50 of Regulation (EU) 2024/1689 (AI Act), fully applicable from 2 August 2026 under its Article 113, requires clearly informing anyone interacting with an AI system when this is not otherwise obvious. The agent includes this notice from the first call or message.
No. The agent integrates with the hotel PMS or restaurant booking engine and with the floor POS/ERP the establishment already has; day-to-day floor, kitchen, stock and billing management stays with Summum Sistemas' hospitality ERP. This page adds the conversational layer for bookings, menu recommendations and demand forecasting on top of those systems.
The module analyses the establishment's own booking history by service (breakfast, lunch, dinner) and day type (weekday, weekend, public holiday) to anticipate the staffing and purchasing needed. It relies on the same Summum IA demand forecasting engine, configured with the specific occupancy and seasonality variables of hospitality rather than the generic setup used for other sectors.
The agent escalates the call or chat to front desk or floor staff, with a summary of the conversation so far, following the escalation conditions set out in the initial blueprint. These conditions can be adjusted once the agent is in production.