Model Context Protocol · Public server
MCP travel connector.
Connect AI assistants to travel search and booking capabilities through the Model Context Protocol.
Public MCP server · Flights & hotels · Booking, not just search
MCP session · example call
Beeline
→ tools/list
- search_flights
- search_hotels
- get_flight_details
- get_hotel_details
- check_flight_availability
- check_room_availability
- hold_flight
- hold_hotel
- get_flight_booking_status
- get_payment_link
→ tools/call search_hotels
{
"destination": "Lisbon",
"check_in": "2026-05-01",
"check_out": "2026-05-04",
"guests": 2
} ← 34 properties · live rates · totals for 3 nights
Straight answer
What is an MCP travel connector?
An MCP travel connector is a connection that allows an AI assistant to interact with travel services through the Model Context Protocol.
- AI assistant
- MCP
- Travel tools
- Travel systems
Tool surface
The tool surface, as the assistant sees it.
Read tools return live data; write tools change the state of a booking.
| Tool | What it does | Type |
|---|---|---|
Flight search search_flights | Availability from connected airline sources, priced for the passenger mix supplied. | Read |
Hotel search search_hotels | Room-level availability for a destination and date range, with live totals. | Read |
Offer detail get_flight_details · get_hotel_details | Fare classes and what each includes; rooms, amenities, policies and photos. | Read |
Availability check check_flight_availability · check_room_availability | Confirms an option is still sellable at the price shown before booking. | Read |
Booking creation hold_flight · hold_hotel · get_payment_link | Holds a selected option, collects traveller details and returns a checkout link. | Write |
Booking retrieval get_booking_status · get_my_bookings | Current state of a booking made through the connector. | Read |
Booking management cancel_flight · cancel_booking · estimate_flight_cancellation | Change and cancellation requests, subject to fare rules and hotel policy. | Write |
Architecture
How Beeline works as an MCP travel connector.
Four layers. The client speaks MCP; supply, contracting and payment sit behind the server. Building and maintaining the underlying travel integrations remains Beeline’s work, not something MCP removes.
- AI applicationLayer 1
- Beeline MCP serverLayer 2
- Beeline travel toolsLayer 3
- Flight & hotel inventory and booking systemsLayer 4
Comparison
Not an API. Not instead of one, either.
| Travel API | MCP travel connector |
|---|---|
| Built for an application you are writing | Built for an AI assistant to discover at runtime |
| You read the docs and write the integration | The assistant reads the tool definitions itself |
| Fixed call sequence in your code | Sequence chosen by the model, turn by turn |
| One integration per application | One server, any client that supports the required protocol features |
| Application → API → travel service | AI assistant → MCP tool → travel service |
Built for
Anyone whose assistant needs to actually book.
- consumer assistants adding travel
- teams building travel agents
- brands with AI ambitions
- internal assistants and corporate travel
- rewards and embedded travel
- assistant surfaces at scale
2.9M hotels 300+ airlines One connector Zero tabs
FAQs · 15 questions
MCP travel connector questions.
MCP and travel
- What is an MCP travel connector?
- An MCP travel connector is a connection that lets a compatible AI application use travel tools through the Model Context Protocol. Depending on the tools exposed by the server, an assistant may be able to search flights or hotels, retrieve booking information, create a booking workflow, or request supported booking-management actions.
- What is the Model Context Protocol (MCP)?
- The Model Context Protocol (MCP) is an open protocol for connecting AI applications to external tools and data sources. MCP defines how clients and servers establish a session and advertise supported capabilities, but an integration works only when the client and server support the required protocol features, transport and authentication.
- What is MCP in travel?
- MCP in travel is the use of the Model Context Protocol to connect AI assistants with travel systems. It allows an assistant to search inventory, retrieve information, and perform actions such as booking or managing travel, rather than only describing travel from memory.
- What is a travel MCP server?
- A travel MCP server exposes travel-related tools that a compatible AI application can discover and call through the Model Context Protocol. The exact tool set depends on the implementation and may include search, offer detail, availability checks, booking creation, booking retrieval or booking-management actions.
- How does an AI travel connector work?
- An AI travel connector works by giving an assistant structured access to travel tools that perform real-world tasks. The assistant discovers the available tools, calls the one that matches the user’s request, and receives live results back — so its answers come from travel systems rather than from the model alone.
- How do AI agents connect to travel APIs?
- AI agents can connect to travel APIs through application-specific integration code or through an MCP server that presents API capabilities as tools the agent can discover and call. MCP standardises the tool interface, but the server provider still has to build, operate, secure and maintain the underlying travel integrations.
- How do AI agents access live travel inventory?
- AI agents access current travel inventory by calling a connected service that queries airline or hotel systems at the time of the request. Beeline can return availability and pricing from its connected sources and recheck a selected option during booking; the result is not final until the booking is confirmed.
Capabilities and boundaries
- Can MCP be used for flight search?
- MCP can be used for flight search when an MCP server exposes a flight-search tool and the client supports the required connection and authentication. A Beeline flight-search integration can accept details such as route, dates and passenger mix and return available offers from Beeline’s connected inventory.
- Can MCP be used for hotel booking?
- MCP can support hotel booking when the server exposes the required hotel-search and booking tools and the client supports the workflow. With Beeline, supported implementations can return hotel and room options, prepare a selected booking and provide a secure checkout link; payment-card details should not pass through the MCP conversation.
- Can any MCP client connect to an MCP travel server?
- Not every MCP client can use every MCP travel server. The client and server must support compatible protocol features, transport and authentication, and the host application must allow the relevant tools and user approvals. The client Beeline supports today is covered in how to add Beeline to Claude.
- What is the difference between an AI travel agent and an MCP travel connector?
- An AI travel agent is the user-facing assistant that helps a traveller plan or book a trip, while an MCP travel connector is the technical connection that gives a compatible assistant access to travel tools. Beeline supplies travel capabilities through MCP, and the client application decides how those tools appear to its users.
- What is the difference between an MCP travel connector and a travel API?
- A travel API exposes endpoints that developers call from application code, while an MCP travel connector exposes tools that a compatible AI application can discover and call at runtime. They are complementary: an MCP server can sit on top of travel APIs and present selected capabilities through MCP.
- Does an MCP connector replace our existing travel API integrations?
- An MCP connector does not replace an existing API integration; it sits alongside it, serving AI clients rather than application code. The same underlying travel systems can be reached both ways.
- Who is an MCP travel connector for?
- An MCP travel connector is for AI assistants adding travel, developers building travel agents, travel companies wanting presence inside AI applications, and enterprise teams adding booking to internal assistants. The common requirement is an AI surface that needs to perform real travel actions.
- How does payment work through an MCP travel connector?
- An MCP travel connector can return a link to a secure hosted checkout so payment happens outside the AI conversation and client application. The checkout should show the final amount, booking terms and payment recipient, and payment-card details should not be sent through MCP tool arguments or chat messages.
Related
/ai-travel-agent
What is an AI travel agent? →
The category page: definition, capabilities, and whether AI can really book.
/claude-travel-agent
AI travel agent in Claude →
How the connector appears to a traveller, and the setup path.
/ai-travel-booking
AI travel booking →
From conversation to booking, and where payment happens.
Put real travel behind your assistant.
Public MCP server · Flights & hotels · Booking, not just search