Reading time: 7 minutes
The guest reserves on the engine, the rates and the payment gateway of the chain itself, which retains direct sales, customer data and the relationship with the guest. Not even the largest chains in the world still have it. Working on ChatGPT and Claude.
Neobookings has solved the most difficult part of the agentic era: having the architecture and infrastructure ready for consumption by agents. We have unified our data into a unique, processed, standardized and cured knowledge base, and we expose it to an open standard, Model Context Protocol (MCP).
Any agent or LLM can consult reliable and structured information, without depending on the interface with which the user interacts. It doesn't matter if it's a chatGPT application, a chatbot or a voice agent: they all access the same source of truth, ready to be consumed from the first day and without redoing the integration for each new wizard.
Hilton, Marriott, Accor, and Radisson compete to dominate reservations in AI assistants, but disrupt the guest's experience and trust
The large international chains are moving to occupy the space of AI attendees, each one by a different path.
Hilton is present on Google AI Mode and on ChatGPT, but when the user decides to book, he redirects it to Hilton.com.
Marriott has chosen another way. Your bet is that the assistant will name your hotels, not to complete the reservation. The chain feeds Google and OpenAI with all possible content to appear in searches, in addition to using the OpenAI advertising platform.
Accor and Radisson Hotel Group have brought their own applications to ChatGPT. Through these, the user looks for group hotels by dates and number of guests, and consults types of rooms, services, location, public rates and exclusive prices for members of their loyalty program.The reservation, however, is then completed on the web pages.
These paths are all necessary, but they share a limit: when the assistant recommends the hotel, it takes the sale out of the conversation. This generates aInterruption of the experience that, in turn, undermines the confidence of the traveler, who does not understand why everything should start from scratch, just at the moment he was making the final decision.
Each jump out of the conversation increases the risk of losing the sale. After years of optimization and digital adjustments to reach a frictionless conversion, getting the guest out of the conversation, especially with the degree of fluidity offered by an LLM, is the greatest of interruptions.
Booking within Google does not resolve direct sales either. In Google AI Mode, the purchase occurs in an interface and a foreign payment system, which is still a showcase shared with the OTAs, which function as majority partners in a bid to appear in the assistant. This drastically decreases the guarantee that a mention will end up becoming a sale.
The complete interface of a booking engine, with photos, map, price calendar and basket, inside ChatGPT and Claude
The guest experience not only has to be smooth and uninterrupted. When we talk about choosing a hotel, the most relevant are the photos and the visualization of the services and dates, even before comparing prices.
With NeoGPT, the guest finds the same interface of the hotel's reservation engine within the AI chat. Receive the results in a fully interactive interface: Map of locations of each hotel in the chain, with its region, zone or amenities selectors. Each result offers the option to open the tab of each hotel, which has its interactive gallery, address, contact, services, reservation terms, price calendar and payment methods.
In this way, choosing a hotel on the hotel's website or doing it in the AI chat offers the experience that the guest expects, with the advantage that now it is a conversation and accompanying the client within the new environment where decisions are made.
Your complete booking engine within ChatGPT and Claude
NeoGPT offers the most powerful (and unique) system on the market. The reservation is completed within ChatGPT or Claude, about the engine, the rates and the payment gateway of the chain itself, which retains direct sales, customer data and the relationship with the guest.
Hotel attendees in ChatGPT and Claude usually limit themselves to talking in plain text, without any appeal -something of maximum importance when it comes to booking a room or hotel experience-, or send the guest to the web to reserve.
NeoGPT offers the same experience as the chain's reserve engine. The reservation begins and ends in the conversation.
NeoGPT within ChatGPT: Bluesea Gran Playa rooms (Mallorca) with photos, regime, rate, remaining availability and discount.
Interactive elements, galleries with filters and map
NeoGPT connects to ChatGPT and Claude via MCP. The guest talks in natural language and the assistant shows you at every step the interface you need: calendar for the dates, selector for adults and children with their ages, rooms with photos, regimen and real price, extras such as breakfast, parking or late check-out, form with the data required The hotel and, in the end, the locator with the abstract.An unlimited range of interactive elements, including galleries with filters and map.
The agent sells with the same visual elements as the booking engine: it shows how many rooms are left in each rate, shows the price crossed out with the discount applied and allows you to select several rooms in the same reservation.Payment is made on the hotel walkway, safely.
Operating today in the LLM, any agent tomorrow
The interface from which the guest reserves will change: today they are ChatGPT and Claude, tomorrow it will be another assistant, a self-employed agent or the voice itself. What determines whether a chain sells in all of them is the actionable infrastructure and data architecture below, which will give agents the ability to check availability, quote, add services, collect, modify and cancel on the real inventory, reliably and at low cost.
Neobookings has built that layer on an open standard, so that the same system can be taken to any wizard who adopts MCP without redoing the chain integration.
At Neobookings we have been pioneers in the intelligent age. We started working with Model Context Protocol a year and a half ago, when the standard was barely used outside of engineering equipment. Today MCP is the road with which the main AI assistants connect to external services.
And this change has already begun. Openai is activating proactive plugin discovery within the conversation itself: ChatGPT can detect when an external capacity is relevant and suggest to the user at the same time. The user no longer needs to know which plugin exists or search for it previously. The wizard can take the ability to the conversation, just when you need it.
That anticipation allows us to reach ChatGPT with NeoGPT and Claude with a product tested in production, with months of learning about how guests talk and reserve.
NeoGPT within Claude.
Today in a touch, tomorrow by voice
NeoGPT supports voice input. The guest can say what they are looking for, review the options on the screen and confirm the reservation without writing or filling in forms, because the voice and the visual interface use the same capabilities of the wizard.
AI attendees are bringing the voice to the center of the experience, and the spoken reserve will be one of the first daily uses. The chain that has its direct channel prepared for that moment will be the one that can capture that demand.
The next step will be for the agent to reserve from start to finish on behalf of the guest, without his having to intervene. A vision that we have from the beginning, a year and a half ago.
In fact, at FITUR 2026, in January, we already projected this scenario. We created a spot to show how we imagined the future of the hotel reservation, with an experience in which the guest could talk with technology and move towards the reserve in an increasingly natural way. What we then proposed as a vision of the future is much closer today to being a reality.