Perhaps, you've heard of Model Context
Protocol (MCP). Maybe you've even done some light research on what it's all
about. But it's still not quite sinking in.
Here are answers to some basic questions about
MCP that will have you nodding with knowledge and understanding instead of
bewilderment at your next big business travel event.
#1: What is Model Context Protocol?
Model Context Protocol is a standard that lets AI models (Claude,
ChatGPT, OpenClaw) connect directly to external data sources and software
systems (airline reservation systems, payment and expense platforms, TMC
booking tools).
Technology experts often reach for physical analogies to explain it, but
just think of it as a common language that's ultimately going to let AI agents
talk to any system.
#2: How is MCP different from an API?
An API is a connection to one service. MCP is a standard that helps AI
find and use many different services (confusingly, often via APIs).
An API also only answers the exact question it's built to answer. To use
an example provided by Travlr ID CEO Gee Mann, if you want a system to be able
to pull passport information from a traveler profile, a specific API needs to
be built to do that and it will only be able to do that one thing.
MCP would let an AI agent work with less specific instructions. You
could ask, "What's missing from this traveler's documentation," and
the agent could scan for the issue and retrieve the passport or any other
information needed.
#3: What's the deal with MCP servers?
An MCP server is a piece of infrastructure
that actually does the translating for this new language. A supplier sets up a
server (either in-house or via a vendor) and the server takes a jumble of
back-end data and puts it into a form AI agents can read.
The server also controls the access to who can
see what, so a negotiated corporate rate for XYZ Company is only surfaced to
XYZ Company travelers.
Sabre launched its own MCP server last
September, and Navan and Vibe have each recently launched MCP servers.
#4: Is an MCP layer something different?
Not exactly. MCP gives AI agents access
to a supplier's data and systems. An MCP layer is a company's version of that
standardized access point. For example, Expensify launched an MCP layer that enables AI
assistants like ChatGPT and Claude to query a client's expense data in plain
language instead of through manual CSV exports.
However, there are also governance and
payment layers that get discussed alongside MCP. A governance layer, according
to AstraNomad's Martijn van der Voort, is what lets agentic systems recognize
complex rules and stop transactions when they run afoul of those rules. A
payment layer leverages (Agentic Commerce Protocol or ACP) and allows AI agents
using MCP to securely authorize and complete financial transactions.
#5: Do I, as a travel manager, need to
worry about this yet?
Not quite. MCP could bring major changes to
corporate travel, impacting booking, content, data, reporting and supplier
relations. But travel technology experts agree it's still early days in terms
of adoption, access and sophistication of the technology.
_____________________________________________________
Note:
This article is based on reporting by Michael Baker. For more, access the full
in-depth piece MCP: Managed
Travel's Next Transformation Catalyst.