NeoGPT, the only system on the market to offer the full booking engine interface and end-to-end reservations within ChatGPT and Claude

Neobookings has solved the hardest part of the agentic era: building the architecture and infrastructure required to make its systems accessible to AI agents. We have consolidated our data into a single knowledge base, processing, normalising and curating it before making it available through an open standard, the Model Context Protocol (MCP).

Any agent or LLM can access reliable, structured information without depending on the interface through which the user interacts. Whether it is a ChatGPT app, a chatbot or a voice agent, they all access the same single source of truth, ready to be used from day one, without having to rebuild the integration for every new assistant.

Hilton, Marriott, Accor and Radisson are competing to own hotel bookings through AI assistants, but their approaches disrupt the guest experience and undermine trust

Major international hotel groups are moving to establish their presence within AI assistants, each taking a different approach.

Hilton is present in Google AI Mode and ChatGPT, but when a user decides to book, they are redirected to their site.

Marriott has chosen a different route. Its strategy is to ensure that the assistant mentions its hotels, rather than completing the booking. The group supplies Google and OpenAI with as much content as possible to increase its visibility in search results, while also using OpenAI’s advertising platform.

Accor and Radisson Hotel Group have brought their own apps into ChatGPT. Through these apps, users can search the groups’ hotels by dates and number of guests, and explore room types, facilities, locations, public rates and exclusive prices for loyalty programme members. The booking itself, however, is completed later on the hotels’ websites.

All these approaches have a role to play, but they share one fundamental limitation: when the assistant recommends a hotel, it takes the transaction outside the conversation. This interrupts the guest experience and, in turn, undermines travellers’ trust. They cannot understand why they have to start from scratch precisely when they are ready to make their final decision.

Every step outside the conversation increases the risk of losing the booking. After years of digital optimisation and refinement aimed at achieving frictionless conversion, taking guests out of the conversation, particularly when LLMs offer such a fluid experience, represents the most significant disruption of all.

Completing a booking within Google does not solve the direct booking challenge either. In Google AI Mode, the transaction takes place through an external interface and payment system. These remain a shared storefront with online travel agencies (OTAs), which act as major partners in the bidding process to appear within the assistant. This significantly reduces the certainty that a hotel mention will ultimately translate into a direct booking.

The full booking engine interface, complete with photos, a map, a price calendar and a basket, within ChatGPT and Claude

The guest experience needs to be seamless and uninterrupted. When choosing a hotel, photographs and the ability to explore facilities and dates visually are among the most important factors, even before comparing prices.

With NeoGPT, guests find the hotel’s actual booking engine interface within the AI chat. Results appear in a fully interactive interface, featuring a map showing the locations of the chain’s hotels, with filters for region, area and amenities. Each result can be opened to display the individual hotel’s profile, complete with an interactive photo gallery, address, contact details, facilities, booking conditions, price calendar and payment methods.

This means that choosing a hotel on its website or within an AI chat delivers the experience guests expect, with the added advantage of a conversational interface that supports customers in the new environment where decisions are being made.

Your complete booking engine within ChatGPT and Claude

NeoGPT offers the most powerful booking system on the market, and the only one to provide this complete functionality. Bookings are completed within ChatGPT or Claude, using the chain’s own booking engine, rates and payment gateway, allowing it to retain direct bookings, customer data and the relationship with its guests.

Hotel assistants within ChatGPT and Claude typically offer little more than plain-text conversations, with none of the visual appeal that is particularly important when booking a hotel room or experience. Alternatively, they redirect guests to a website to complete the booking.

Watch the full booking process.

NeoGPT delivers the same experience as the chain’s own booking engine. The booking starts and finishes within the conversation.

Interactive elements, filterable galleries and maps

NeoGPT connects to ChatGPT and Claude through MCP. Guests interact using natural language, and the assistant displays the interface they need at each stage: a date calendar; an adult and child selector, including children's ages; rooms with photos, board basis and actual prices; extras such as breakfast, parking and late check-out; a form collecting the information required by the hotel; and, finally, the booking reference and reservation summary. The system supports an unlimited range of interactive elements, including filterable galleries and maps.

The agent sells using the same visual elements as the booking engine. It shows how many rooms remain available at each rate, displays the original price crossed out alongside the applied discount, and allows guests to select multiple rooms within a single booking. Payment is processed securely through the hotel’s own payment gateway.

Working with today's LLMs, ready for any agent tomorrow

The interface through which guests make their bookings will continue to evolve. Today, it is ChatGPT and Claude; tomorrow, it could be another assistant, an autonomous agent or voice itself. What determines whether a hotel chain can sell through all these channels is the underlying infrastructure and actionable data architecture. This is what will enable agents to check availability, quote prices, add services, take payments, modify bookings and process cancellations against real inventory, reliably and at low cost.

Neobookings has built this layer on an open standard, allowing the same system to be deployed in any assistant that adopts MCP, without rebuilding the integration with the hotel chain.

At Neobookings, we have been pioneers in the agentic era. We began working with the Model Context Protocol a year and a half ago, when the standard was barely used beyond engineering teams. Today, MCP is the mechanism through which leading AI assistants connect to external services.

And this shift has already begun. OpenAI is rolling out proactive plugin discovery within the conversation itself. ChatGPT can identify when an external capability is relevant and suggest it to the user at that moment. Users no longer need to know which plugin exists or search for it in advance. The assistant can bring that capability directly into the conversation, precisely when it is needed.

This head start means that NeoGPT is now available in ChatGPT and Claude as a product that has already been tested in production, backed by months of learning about how guests interact with AI assistants and make bookings.

Today, through touch. Tomorrow, through voice

NeoGPT supports voice input. Guests can describe what they are looking for, review the options on screen and confirm their booking without typing or filling in forms, because voice and the visual interface draw on the same assistant capabilities.

AI assistants are bringing voice to the centre of the user experience, and conversational booking will be one of its first everyday applications. The hotel chain that has its direct channel ready for this shift will be the one best positioned to capture that demand.

The next step will be for the agent to complete the entire booking on the guest’s behalf, without requiring any intervention from them. This is a vision we have pursued from the outset, a year and a half ago.