Key Takeaways
- Travelport, one of the three global GDS operators, is rebuilding its travel retailing and distribution stack with Anthropic's Claude and Cognizant's engineering muscle, starting with Trip Services and an MCP-based conversational booking layer.
- The structural gap the partnership targets is the one between AI agents that can plan trips and transactional systems that can actually book them — and Anthropic, the organization that invented MCP, has been pulled in to close it.
- For e-commerce operators, the rebuild of travel retailing is a leading reference case for any business that sells against complex inventory, dynamic pricing, and ongoing servicing — including subscriptions, ticketing, and experiential products.
Travelport puts Anthropic and Cognizant at the center of its AI rebuild
Travelport partners with Cognizant and Anthropic to deploy Claude across travel retailing and distribution platforms.
www.businesstravelnewseurope.comOn May 27, 2026, Travelport, one of the three global GDS operators alongside Amadeus and Sabre, announced a three-way collaboration with Anthropic and Cognizant. According to Cognizant's release, Claude will be embedded into Cognizant's engineering stack to modernize how Travelport builds, tests and maintains software across its travel retailing and distribution platforms. Business Travel News Europe reported the same day that "Cognizant is deploying Anthropic's AI assistant Claude to modernise Travelport's technology approach."
The shape of the deal does not come out of nowhere. Anthropic announced in November 2025 that Cognizant would make Claude available to up to 350,000 employees, placing Anthropic at the core of Cognizant's "AI Builder" model. The Travelport engagement is the first large vertical-specific use case to come out of that strategic relationship, and the choice of travel — rather than the financial services vertical Cognizant initially mentioned — is itself a signal.
Cognizant CEO Ravi Kumar S framed the work as "giving Travelport the tools to move faster and deliver higher quality at scale to meet the challenge of a changing travel distribution landscape." This is not described as a limited pilot. It is positioned as a long-running transformation that pushes Claude into the day-to-day engineering surface of how Travelport ships software.
Trip Services automation and "AI that closes the booking"
The collaboration's first focus is Travelport Trip Services, the cloud-native API platform that handles bookings, exchanges, refunds and disruption servicing. This is the spine of travel retailing, and API-driven transactions through Trip Services have grown from 43% of all Travelport bookings in 2022 to 63% today, according to Travel Distribution News.
Cognizant integrates Claude into code development, test creation and pull-request review, then uses Claude's large context window to read across Travelport's codebases and surface the business logic embedded inside them. Reasoning across large, complex codebases is, as Rich O'Connell of Anthropic put it, "exactly what travel infrastructure demands" — and it is one of the technically hardest pieces of enterprise modernization.
Internal modernization is only half the picture, though. The other half is the customer-facing surface of Trip Services. The release calls out automated exchanges and disruption intelligence as the first capabilities to reach market, with launches expected later this year. The concrete picture is an agent managing a business traveler instantly surfacing routes with statistically lower disruption risk. Travelport claims that saving even one hour per agent per day across a large TMC translates into millions of dollars in annual productivity improvement.
MCP wires "conversation to confirmed booking" into one loop
The most strategically loaded part of the announcement is that Travelport is committing to MCP (Model Context Protocol) for its upper interface layer. CEO John Mangelaars said: "Anthropic developed MCP, the protocol that lets AI agents interact directly with external systems and data. Choosing the organization that invented that protocol was a straightforward decision." That is more than vendor selection language — it is a statement about which standard will govern the next decade of travel agentic commerce.
MCP has spread quickly as the open standard for connecting AI agents to external systems and data. Skift's end-of-2025 explainer, MCP Explained: The AI Standard Reshaping Travel Tech, notes that MCP has been adopted by OpenAI, Google, Microsoft, and Amazon and is now stewarded by the Linux Foundation. On the travel side, Amadeus, BCD Travel, Booking.com, and Expedia have all moved on MCP, turning it into the common substrate for AI agents that need to query travel data and execute bookings.
MCP on its own, however, cannot run travel retailing. It is not designed to handle the full shopping, booking, and servicing workflow. Travelport's design puts Trip Services' cloud-native transactional fabric underneath and the MCP-based conversational interface on top, translating a traveler's natural-language intent directly into a confirmed booking against live inventory. In the words of the release, it is "intelligent infrastructure designed to close the gap between AI-driven travel intent and a confirmed booking."
Anthropic also developed MCP, the protocol that lets AI agents interact directly with external systems and data. Choosing the organization that invented that protocol was a straightforward decision. Their approach to safety, reliability and controllability matters as much: travel is a high-trust environment where data is sensitive and the consequences of errors are real.
Reading the three-way division of labor
Agentic commerce debates tend to fixate on "which AI does the buying." Travelport's setup is a cleaner illustration that the next-generation operational stack is being built as distinct layers, each owned by a different player.
| Player | Role | What they bring |
|---|---|---|
| Travelport | Travel infrastructure and distribution | Trip Services (booking, exchanges, refunds, disruption servicing), MCP-based upper interface layer, distribution reach across 165+ countries |
| Anthropic | Foundation AI and protocol | Claude models, large context window for codebase reasoning, originator of the Model Context Protocol (MCP) |
| Cognizant | Engineering and delivery | Claude-embedded engineering platform for code, test and PR review, multi-agent stack including Neuro-san, AI Builder delivery model |
Cognizant's AI Builder strategy explicitly takes on industry context, end-to-end systems integration, and day-to-day operational accountability, with the goal of moving enterprises from AI experimentation to production at scale. Anthropic supplies the foundation models and the protocol that is becoming the industry standard. Travelport contributes a distribution network across 165 countries plus the messy, real-world inventory and operations data that agents have to act on. Drop any one of the three, and travel retailing's AI rebuild does not close.
What makes the pattern interesting beyond travel is that this three-way shape is a transferable template. Cognizant's November 2025 announcement framed the Anthropic partnership as starting with financial services; travel is the next vertical, and it is the first that involves complex inventory plus dynamic pricing plus continuous servicing. The same template extends naturally to ticketing, subscriptions, and experiential products.
Redefining the GDS, and travel's AI transformation
Travelport has spent the last several years recasting itself from "GDS operator" to "intelligent infrastructure layer for AI-driven travel commerce." Travel Distribution News's March 2026 analysis — coming on the back of $50 million in fresh shareholder investment, 12% EBITDA growth, and the consolidation of operations onto Trip Services — concluded that "the GDS era may not be ending. It may be transforming into something else entirely. And Travelport appears to be the first of the three major players to make that bet explicitly."
The bet is sharpening now because the surrounding landscape has moved. Booking.com and Expedia are building proprietary AI experiences, Amadeus has staked out its own MCP/UCP-anchored direction, and TMCs from FCM (Sam) to Navan (Vibe) are shipping AI ecosystems of their own. Travelport, sitting in the infrastructure layer, faces a real risk of being routed around if it cannot keep pace with where AI agents want to consume travel content.
What works in its favor is the multi-year investment that has already produced a cloud-native Trip Services and a normalization layer that spans EDIFACT, NDC, low-cost carrier content, and hotel supply. AI agents cannot operate on fragmented, inconsistent data — they need structured, enriched content they can query and act on without human interpretation. Pulling in Anthropic and Cognizant to rebuild the software delivery process is the move that raises the throughput of new capabilities on top of that foundation.
Three takeaways for e-commerce operators
Travel is running ahead of e-commerce on several fronts that the rest of digital retail will eventually hit. Complex inventory, dynamic pricing, mid-flight changes and cancellations, subscription-style customer servicing — the issues travel retailing is solving for now are largely the issues broader commerce will be solving for next.
The first takeaway is the single line from conversational intent to confirmed transaction. AI-driven product discovery is widely deployed, but landing on an actual purchase against live inventory still requires product data, inventory APIs, tokenized payments, authorization, and post-purchase servicing to be threaded into one continuous path. The MCP-based interface layer Travelport is building runs on the same design thinking as the moves Shopify Storefront and Adobe Commerce are making in storefront-side MCP for agentic commerce.
Second, modernization starts in the engineering stack, not the consumer UI. The first place Claude is being pointed in this partnership is Cognizant's code development, test and PR review workflow. E-commerce operators preparing for agentic commerce should expect the same — the leverage shows up first in your ability to read and rebuild your own systems, and only later in customer-facing features.
Third, the model-provider plus integrator plus industry-platform shape of the partnership is itself a useful planning frame. Anthropic owns model and protocol, Cognizant owns delivery, Travelport owns the industry substrate. Operators trying to figure out how to build their own AI commerce stack will get to production faster by being explicit about who they want filling each of those layers than by trying to assemble everything internally.
Bottom line
The Travelport-Anthropic-Cognizant collaboration is both a rebuild of travel retailing around MCP and the first large vertical proof point for the AI Builder delivery model. Visible outcomes — Trip Services automation, disruption intelligence, productivity gains for TMCs — are slated to ship this year.
The signals worth watching from here are concrete: how much real-world transaction volume the first customer-facing Trip Services capabilities pick up, how quickly MCP-based conversational booking spreads across other OTAs and TMCs, and how Cognizant's AI Builder model lands in verticals beyond travel. The shift from AI experiments to AI rebuilds of industry infrastructure is now visibly underway.
