PayPal's Agentic Commerce Push into Travel: The Payments Layer and Five Infrastructure Paths Behind the Whitepaper

A strategy document from a payments company, addressed to travel

Agentic commerce in travel: Preparing for the industry's next big shift

The report provides a look at how agentic AI is reshaping travel commerce and what travel merchants must do to prepare for the next phase of AI-driven discovery, booking and payment.

www.phocuswire.com

In April 2026, PhocusWire published a whitepaper sponsored by PayPal, titled "Agentic Commerce in Travel: Preparing for the Industry's Next Big Shift." Written by Michelle Bruno, the 13-page report works through the state of agentic commerce in travel, the obstacles to the transition, the infrastructure options, and the preparation merchants should undertake.

Agentic commerce refers to transactions in which AI agents research, negotiate, and complete purchases on a user's behalf. The report frames the shift as a progression along a single line — "AI in travel exists on a continuum" — from generative AI that supports research and planning, to agentic AI that executes tasks such as booking flights and completing payments, to agentic commerce built on those capabilities. It also notes that the degree of autonomy is a design choice, ranging from human-in-the-loop confirmation to full autonomy.

Reading this only as travel-industry commentary would miss the point. What matters is that the party behind the report is a payments infrastructure company. Since the fall of 2025, PayPal has made a sequence of moves to build the payments and catalog plumbing between AI channels and merchants. Travel is the next vertical that strategy has reached, and the report's analysis and recommendations bear directly on the design decisions of any business that holds product data and processes payments.

56% of travelers already plan with AI: the data behind the urgency

The case for urgency opens with consumer data. According to Phocuswright's U.S. study "The AI Surge: Travel's Fastest Behavioral Shift in a Decade," 56% of U.S. travelers — one in two — used ChatGPT, Google's AI Mode, or similar tools for trip planning or in-destination assistance over the prior 12 months. Adoption is spreading in Europe as well: as of the second quarter of 2025, 22% of travelers in the UK, 19% in France, and 16% in Germany used generative AI for travel.

The supply side is moving in parallel. Phocuswright's enterprise study "Budgets, Barriers and the Race to Agentic AI" finds that 83% of surveyed travel companies already use generative AI in their operations, and around six in ten are experimenting with or scaling agentic AI. A similar share is exploring interoperability protocols such as MCP (Model Context Protocol), which connects AI to external tools, and A2A (Agent-to-Agent), which links agents to one another.

The connection paths themselves are changing too. The report points to WebMCP, which Google announced as available in early February 2026. Through a single API, navigator.modelContext, AI agents can interact directly with a website's backend functions without operating its graphical interface — a channel the report calls a hidden highway. On airline, hotel, and OTA sites, a route is opening where bookings complete without a single click.

The scale of the AI surfaces on the receiving end is hard to ignore. The monthly active user figures the report cites are 2.8 billion for ChatGPT, 450 million for Gemini, 30 million for Copilot, 22 million for Perplexity, and 20 million for Claude. If product discovery migrates to these conversational surfaces, merchants need circuitry that connects real-time inventory and payments to them. That premise runs through the entire report.

Wallet, Agent Ready, ACP: PayPal's moves in sequence

The fastest way to understand the report's recommendations is to line up PayPal's own moves in order. The starting point is the October 28, 2025 launch of Agentic Commerce Services. The suite rests on two pillars: Agent Ready, which lets existing PayPal merchants accept payments on AI surfaces without additional development work — with fraud detection, buyer protection, and dispute resolution built in, and availability slated for early 2026 — and Store Sync, which connects product catalogs, inventory, and fulfillment data to AI channels, with product discovery on Perplexity planned before the end of 2025.

The same day, PayPal announced a partnership with OpenAI. PayPal's wallet joins ChatGPT's Instant Checkout, and PayPal adopts ACP (Agentic Commerce Protocol), the open-source commerce specification published by OpenAI. In 2026, the deal is set to bring the product catalogs of PayPal's merchant network into commerce on ChatGPT.

The extension into travel came on February 12, 2026, with the announced partnership with Sabre and Mindtrip. Mindtrip's conversational AI takes in the traveler's preferences, Sabre's APIs — covering more than 420 airlines and over 2 million lodging properties — handle search, pricing, availability, and booking, and PayPal provides identity verification and in-chat checkout. At announcement, the companies pointed to a launch in the second quarter of 2026, starting with flights and expanding to hotels in phases. The whitepaper presents this effort as a live example of a standalone agentic AI travel assistant. We cover how the system works in a separate article.

The report itself devotes a chapter to retail's head start, citing in-chat purchases in Perplexity's paid tier, checkout inside Microsoft Copilot, and the surge in generative-AI-referred traffic Adobe observed (up 1,300% year over year in November-December 2024), arguing that travel is positioned to retrace retail's successes. Lined up this way, the document's role becomes clear: it uses market data and an implementation framework to substantiate that PayPal's strategy of owning the transaction layer between AI channels and merchants — anchored in payments and identity verification — has now reached travel, a vertical far more complex than retail.

Five infrastructure options, and how to read a sponsored report

For practitioners, the report's core is its taxonomy of five ways to stand up agentic commerce infrastructure, told through a house-building analogy. Building in-house from the ground up; building on hyperscaler platforms from Google, Amazon, or Microsoft; horizontal agent frameworks, which amount to buying the tools and blueprints; best-of-breed point solutions that combine different vendors per function; and end-to-end agentic commerce platforms that hand the whole build to one contractor.

The weight of the prose clearly favors the fifth option. Its advantages get the most careful treatment: a single platform spanning discovery, booking, payment, and post-purchase servicing; support for multiple commerce protocols; and customer data owned and controlled by the merchant rather than an intermediary. Since that territory overlaps squarely with what PayPal sells, it is healthy to discount the recommendation as one made in a sponsored report.

Even discounted, the comparison axes hold up. Degree of control versus cost, speed to launch, fit with industry-specific workflows, and access to talent — the optimal answer shifts depending on which axis a business prioritizes. Given Gartner's projection that over 40% of agentic AI projects will be canceled by the end of 2027 for lack of careful, strategic decisions, a framework for comparing the options structurally has value in its own right.

Trust, regulation, and edge cases: what stands in the way

The report is specific about the friction points. Trust and control come first. Agents spend money, change bookings, and consume loyalty points. Who bears responsibility for a botched transaction, and how are customers compensated? eBay's revision of its terms of service to bar AI agents and chatbots from connecting without consent shows that the question of how much autonomy to grant remains unsettled across industries.

Regulatory and accounting uncertainty persists as well. In travel, where consumer protection law, payment regulation, and tax and reporting requirements overlap, it is not even clear what can legally be automated. The industry is also dense with edge cases — interline tickets spanning multiple airlines with mixed fares, multi-currency refunds, group bookings with partial refunds — where inconsistent data or policies translate directly into failures.

That is why the report does not advocate jumping straight to full autonomy. It positions existing AI travel assistants such as Delta's Concierge, Expedia's Romie, and Marriott's RENAI as precursors, and treats pilot programs and first-generation AI apps — rolled out to customers incrementally while accuracy improves — as the realistic path. The same sequencing applies unchanged to physical e-commerce, which carries its own heavy load of post-purchase exceptions in returns, exchanges, and warranties.

Start with a catalog audit: the merchant readiness list

Toward the end, the report lays out the six preparation steps PayPal recommends to merchants: audit the product catalog and data structure, choose an integration path, build and deploy commerce API infrastructure, negotiate commercial agreements with each AI platform, test and monitor, and maintain protocol compliance.

Placing the catalog audit first is telling. Clear product titles, comprehensive descriptions, accurate attributes, rich metadata — agentic AI surfaces, the report notes, prioritize brands with well-structured catalogs. The pattern in which product data readiness determines discoverability holds whether the product is a hotel room or a physical SKU.

Equally notable is the observation that agreements and protocol compliance are ongoing work. Merchants that choose direct integration must negotiate data usage rights and fraud liability with each AI platform individually, and must keep pace as commerce protocols such as ACP evolve. This is not a one-time connection but an operating commitment, and the question is whether a business can sustain it.

What justifies the effort, in the report's telling, is the size of the opportunity. McKinsey projects that by 2030 the U.S. B2C retail market alone could see up to $1 trillion in revenue flowing through agentic commerce, with global projections reaching $3 trillion to $5 trillion. IATA projects airline ancillary revenue — fees beyond the fare, such as seat selection and baggage — at $144 billion for 2025, and the report points to contextual, AI-driven offers as a lever that could lift that line further.

Conclusion

The report closes with a warning from Lareina Yee, senior partner and director of technology research at McKinsey Global Institute:

"

Before long, nearly all retailers will have to grapple with the fact that a significant percentage of their customers will not be human users but rather AI agents. The challenge will be to get out in front of it now, before your rivals do.

A meaningful share of customers will soon be AI agents rather than humans, and the test is whether a business gets ahead of that shift before its rivals. What this whitepaper reveals is an infrastructure company that controls payments and identity verification moving to set the standard for the transaction layer, with travel — one of commerce's most complex verticals — as the stage. Seen from the merchant's side, it is a tug-of-war between launching faster on external infrastructure and keeping control of customer relationships and data. The axes for comparing the five infrastructure options, and the readiness list that begins with a catalog audit: these two tools travel well beyond travel, to every retail and e-commerce operator preparing to connect products and payments to AI channels.