NDC gave airlines a structured distribution data backbone. However, a new standard—Model Context Protocol (MCP)—could make that backbone far more usable for AI.
Airlines have invested in NDC for almost a decade. As a result, many airlines can now create structured offers and orders for internet-based distribution. Even so, a key question remains: are airline websites and APIs ready for LLM-based travel assistants?
The reality of airline data exchange today
Here’s what most third-party integrations still face:
1) Fragmented NDC implementations
Every airline “interprets” NDC a bit differently. As a result, basic interoperability remains challenging.
2) Legacy integration patterns
Many NDC stacks still rely on XML/SOAP. As a result, integrations can be slower and heavier than modern JSON API approaches.
3) Flat adoption of new capabilities
In many cases, NDC flows replicate older EDIFACT processes. Therefore, personalization and true retailing benefits are not fully unlocked.
Why MCP changes the game for AI agents
This is where Model Context Protocol (MCP) comes in. MCP is designed for AI agents, and it works like a universal adapter.
Instead of building a custom integration for every tool and workflow, MCP provides a standard bridge. In other words, you can connect AI agents to airline capabilities through MCP connectors.
As a result, travel-specific agents can do tasks such as:
-
Search flights and prices
-
Reshop existing orders
-
Upsell ancillaries
-
Trigger changes, refunds, or servicing actions
Proof point
Kiwi.com has already built an MCP server where a chatbot can search flights and return instant booking links. That’s a practical example of what “agent-ready” distribution can look like.
What the future airline distribution can look like
NDC provides the structured offers and orders. Meanwhile, MCP can provide the agent-friendly connector layer. Together, they can support real-time retailing in an AI-driven travel search experience.
How Threedot can help
At Threedot, we help airlines:
-
Benchmark NDC maturity
-
Map which flows are agent-ready
-
Design MCP-enabled AI workflows that drive conversion, not just compliance
To discuss more about whether your distribution stack is AI agent-ready, please reach out to hello@threedot.io
For more updates, please follow Threedot
