伙伴关系
旅游港和意识到扩大AI伙伴关系，将旅游规划与预订联系起来
在 Google 上的首选来源中添加 Unite.AI
旅游港和认识公司扩大了它们的战略伙伴关系，在旅游零售基础设施、软件工程和业务工作流程中嵌入AI。 宣布 10月27日，该倡议以20年的关系为基础，并针对一项实际挑战：将旅行者的要求变为已完成的交易。
根据公告，第一个现代化建设阶段已经完成。 AI工具目前是推出，用于代码生成，测试创建和拉出请求审查，最初侧重于Travelport TripServices，支持预订，交换，退款，服务。 认识小组将与旅行港合作，跨越18个国家。
规划和交易之间的区别是该项目的核心。 一个对话系统可以建议出行，但完成它需要可用的存货，一致的定价信息，以及能够执行请求的预订服务。 旅游港现有产品显示必须在哪里建立这种联系。
AI授权预订背后的基础设施
旅行港经营一个全球分销系统，将旅行供应商与各机构、在线旅行销售商和旅行管理公司联系起来。 其Travelport+平台将来自航空公司，酒店，汽车租赁公司等来源的内容与搜索，预订，服务工具相融合。 机构可以通过Smartpoint应用程序工作或通过API集成。
该内容通过不同渠道运出，包括传统发售格式和航空公司使用"新发售能力"的API，或NDC. 在AI系统能够有意义地比较报价之前，这些不同的表述需要足够一致，以便搜索和采取行动。 否则，令人信服的建议可能掩盖实际列入内容或可记录内容的差异。
旅游港的 " 内容规范层 " 通过使报价正常化、丰富和去化来解决部分问题。 其内容优化器新增是一个机构控制的规则引擎，用于决定哪些传统空气和NDC提供，以及如何对其进行优先排序。 这些是现有的零售能力，为更广泛的AI程序提供了超越对话界面的基础。
公司于2026年6月正式将发布 TripServices作为云-内生API平台。 Travelport将其描述为航空，酒店和辅助内容的单一连接，并配有嵌入式的管弦乐，用于预订工作流程和排列报价的机器学习模型。 它还为管理变化、取消和中断提供了工具。
对于构建AI出行服务的开发者来说，建筑吸引力是直截了当的：模型可以解释请求，而基础平台则处理结构化搜索和交易工作流程。 可靠的API提供了一个系统建议与其可以执行的更明确的界限。
将软件交付与产品一起现代化
扩大的伙伴关系涵盖三个相关领域：AI辅助工程、工作流程自动化和AI本地产品能力。 称 John Mangelaars, Travelport的首席执行官。
奠基时间在此公告之前。 在5月与Anthropic的合作中，公司描述了通过基于"模型背景协议"(Model Context Protocol)的架构，将Claude跨越软件开发，并将对话出行请求与现场可用性联系起来的计划。 先前的公告还概述了自动交换、重新预订和破坏情报的野心。
MCP为AI应用程序提供了一种与外部工具和数据互动的方式。 在这种背景下，它的有用性取决于这些工具背后的预订服务：理解一个请求和引用一个操作是单独的步骤，操作仍然需要产生一个有效的结果。 5月的计划描述了生态系统的预定方向，而不是证明已经具备了所有拟议的旅行者特征。
认知的AI构建者方法强调围绕企业工作流程、组织背景、治理和可审计性进行定制工程。 这在旅行中很重要，在旅行中，一个有用的人工智能系统必须与一个机构已经依赖的系统和政策合作，并使需要人类判断的案件升级。
其更广泛的应用服务组合包括用于AI辅助软件交付的流程源和用于从实验到部署管理agent系统的神经AI Engine. 这说明认识到了更广泛的技术能力；旅行港公告没有将这两个平台确定为这种参与的具体组成部分。
服务意志测试伙伴关系的价值
旅游港10月26日在商务旅行中讨论AI时，交易、退款以及中断的行程将时间压力与客户偏好、公司政策和业务限制结合起来。 更快地获得相关信息有助于顾问，但当变革产生实际后果时，问责制仍然至关重要。
现有工具已经涉及这项工作的部分内容。 Travelport描述生产力自动化器是创建重复任务规则的一种方式，而其MyTravelport助手帮助用户找到支持和业务信息。 认识和Anthropic合作正在探索AI如何承担更多的人工服务工作量。
这一重点也反映了Unite.AI最近接受CIUSTERY的Babak Hodjat采访时所讨论的企业部署挑战：随着agents进入商业系统，互操作性、高效协调和信任变得更加重要。
对各机构来说，有意义的成果将是减少调和报价和服务预订的工作，在旅行者需要改变时可以可靠地执行。 扩大的伙伴关系将产品雄心与交付所需的工程工作联系起来。
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Partnerships
Travelport and Cognizant Expand AI Partnership to Connect Travel Planning With Bookings
Add Unite.AI to your preferred sources on Google
Travelport and Cognizant have expanded their strategic partnership to embed AI across travel retailing infrastructure, software engineering, and operational workflows. Announced October 27, the initiative builds on a 20-year relationship and targets a practical challenge: turning travelers’ requests into completed transactions.
According to the announcement, the first modernization phase is complete. AI tools are now being introduced for code generation, test creation, and pull-request review, initially focusing on Travelport TripServices, which supports bookings, exchanges, refunds, and servicing. Cognizant teams will work with Travelport across 18 countries.
The distinction between planning and transacting is central to the project. A conversational system can suggest a trip, but completing it requires usable inventory, consistent pricing information, and booking services that can carry out the request. Travelport’s existing products show where that connection has to happen.
The Infrastructure Behind an AI-Powered Booking
Travelport operates a global distribution system connecting travel suppliers with agencies, online travel sellers, and travel management companies. Its Travelport+ platform combines content from airlines, hotels, car rental companies, and other sources with search, booking, and servicing tools. Agencies can work through Smartpoint applications or integrate through APIs.
That content arrives through different channels, including traditional distribution formats and airline APIs using New Distribution Capability, or NDC. Before an AI system can compare offers meaningfully, those different representations need to become consistent enough to search and act on. Otherwise, a convincing recommendation can conceal differences in what is actually included or bookable.
Travelport’s Content Curation Layer addresses part of this problem by normalizing, enriching, and deduplicating offers. Its Content Optimizer adds an agency-controlled rules engine for deciding which traditional air and NDC offers appear and how they are prioritized. These are existing retailing capabilities that give the broader AI program a foundation beyond a conversational interface.
The company formally launched TripServices in June 2026 as a cloud-native API platform. Travelport describes it as a single connection for air, hotel, and ancillary content, with embedded orchestration for booking workflows and machine-learning models that rank offers. It also provides tools for managing changes, cancellations, and disruptions.
For developers building an AI travel service, the architectural appeal is straightforward: the model can interpret the request while the underlying platform handles structured searches and transaction workflows. Reliable APIs provide a clearer boundary between what a system proposes and what it can execute.
Modernizing Software Delivery Alongside the Product
The expanded partnership covers three connected areas: AI-assisted engineering, workflow automation, and AI-native product capabilities. “What excites me is how much AI can change what our platform does for customers,” said John Mangelaars, Travelport’s CEO.
The groundwork predates this announcement. In their May collaboration with Anthropic, the companies described plans to use Claude across software development and to connect conversational travel requests with live availability through an architecture based on the Model Context Protocol, or MCP. That earlier announcement also outlined ambitions for automated exchanges, rebooking, and disruption intelligence.
MCP provides a way for an AI application to interact with external tools and data. In this setting, its usefulness depends on the booking services behind those tools: understanding a request and invoking an operation are separate steps, and the operation still needs to produce a valid result. The May plans describe the intended direction of the ecosystem, rather than proof that every proposed traveler-facing feature is already available.
Cognizant’s AI Builder approach emphasizes custom engineering around enterprise workflows, organizational context, governance, and auditability. That matters in travel, where a useful AI system must work with the systems and policies an agency already relies on, and escalate cases that need human judgment.
Its broader application-services portfolio includes Flowsource for AI-assisted software delivery and Neuro AI Engineering for managing agent systems from experimentation through deployment. These illustrate Cognizant’s wider technology capabilities; the Travelport announcement does not identify either platform as a specific component of this engagement.
Servicing Will Test the Partnership’s Value
Travelport’s own October 26 discussion of AI in business travel places particular emphasis on what happens after a reservation is made. Exchanges, refunds, and disrupted itineraries combine time pressure with customer preferences, corporate policies, and operational constraints. Faster access to relevant information can help a consultant, but accountability remains essential when a change has real consequences.
Existing tools already address parts of that work. Travelport describes Productivity Automator as a way to create rules for repetitive tasks, while its MyTravelport assistant helps users find support and operational information. The Cognizant and Anthropic collaboration is exploring how AI can take on more of the manual servicing workload.
That focus also echoes the enterprise deployment challenges discussed in Unite.AI’s recent interview with Cognizant’s Babak Hodjat: interoperability, efficient coordination, and trust become more consequential as agents gain access to business systems.
For agencies, the meaningful outcome will be less effort spent reconciling offers and servicing reservations, with dependable execution when a traveler needs a change. The expanded partnership links that product ambition to the engineering work required to deliver it.
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.