AI agents are reshaping how the world books travel, and the window to act is two years. Inside: the $1.72T market shift, the trust gap, and what OTAs must build to survive.
March 3, 2026
17
min read
Instructions
If you intend to use this component with Finsweet's Table of Contents attributes follow these steps:
Remove the current class from the content27_link item as Webflows native current state will automatically be applied.
To add interactions which automatically expand and collapse sections in the table of contents select the content27_h-trigger element, add an element trigger and select Mouse click (tap)
For the 1st click select the custom animation Content 27 table of contents [Expand] and for the 2nd click select the custom animation Content 27 table of contents [Collapse].
In the Trigger Settings, deselect all checkboxes other than Desktop and above. This disables the interaction on tablet and below to prevent bugs when scrolling.
Agentic Travel: The Distribution Shift Reshaping How the World Books (2026)
Industry Report - February 2026 - yugo.finance
Agentic Travel
The Distribution Shift That Will Reshape How the World Books
20 min read·February 2026·By Yugo Research
AI agents that autonomously search, compare, book, pay for, and manage travel on behalf of consumers represent the most consequential distribution shift since OTAs displaced traditional travel agencies in the 2000s. The window to act is two years.
McKinsey projects agentic commerce to reach $3-5 trillion globally by 2030. GenAI traffic to travel sites grew 4,700% year-over-year by mid-2025. Yet only 2% of travelers currently allow an AI to complete a booking without oversight. The gap between intent and action is the market to win - and the companies that close it will own the next era of travel distribution.
01 - Scale and Velocity
A $1 Trillion Market Being Rearchitected in Real Time
Global travel gross bookings reached $1.72 trillion in 2025, with online channels accounting for $1.07 trillion and projected to hit $1.2 trillion by 2026. The agentic commerce overlay is materialising faster than most executives anticipated.
$1.72T
Global travel gross bookings 2025
4,700%
GenAI traffic growth to travel sites YoY mid-2025
50%+
Travelers ready for full AI trip booking (Accenture)
$3-5T
McKinsey agentic commerce projection by 2030
Consumer readiness is split along a fault line that defines the strategic imperative. Accenture's 18,000-person survey found over 50% of travelers are ready for AI agents to plan and book entire trips. Forrester reports 53% of Gen Z consumers would delegate travel booking to an agent. BCG found 70% of leisure travelers under 35 already use AI for travel planning. But Skift's definitive finding anchors the reality: just 2% currently allow AI to complete a booking without oversight.
McKinsey projects up to $1 trillion in U.S. B2C retail revenue orchestrated by agentic commerce by 2030 - 3-5x the current online travel market.
Travel sits at a unique intersection of vulnerability and opportunity compared to other verticals. Healthcare leads AI agent adoption at 68%, and financial services is projected to invest $97 billion in AI by 2027. But travel is arguably more exposed than any other vertical: the product is entirely digital, inherently high-frequency, and the distribution stack is already dominated by intermediaries. McKinsey notes that AI-generated recommendations achieve 4.4x higher conversion rates than traditional search, and shoppers arriving via AI chatbots are 38% more likely to purchase.
02 - Who Is Building What
Competitive Landscape as of Early 2026
The agentic travel map divides into clear tiers. Understanding where each player sits determines how quickly the window compresses for everyone else.
Full-stack agentic deployments in production. Navan's Ava virtual agent handles 8,000+ chats daily, resolves 50%+ without human intervention, and claims zero critical hallucinations in production - a benchmark no competitor has matched. Priceline's Penny was the first OTA chatbot to enable full in-chat booking. Booking Holdings commands the most comprehensive position, with CTO Rob Francis describing AI as "the operating fabric" of the company.
Strong Contenders
Expedia - Tripadvisor - Hilton - Hopper - Sabre
Distribution breadth strategy: ChatGPT app, Perplexity Comet, Microsoft Copilot, Operator integrations. Expedia CEO Ariane Gorin is publicly aggressive on AI strategy, though the flagship Romie assistant remains in beta. Hilton reports a 50% boost in direct bookings attributed to AI implementation and AI dynamic pricing boosting revenues 5-8%.
Infrastructure-layer bets: building the rails all agents will call. Kiwi.com was among the first pure MCP server deployments in travel. Sabre's PayPal-Mindtrip partnership (February 2026) is billed as the "industry's first end-to-end agentic AI travel experience," combining GDS inventory, payment infrastructure, and AI planning for a Q2 2026 launch.
At Risk
Marriott - IHG - Amex GBT / SAP Concur - United Airlines - Accor
Both Hilton and Marriott filed SEC disclosures in February 2026 explicitly warning that AI platforms could shift bookings toward commission-charging OTAs. Marriott's AI chatbot remains at prototype stage, previewed to just 75 loyalty members. CEO Anthony Capuano's admission that "we haven't even put on our uniforms yet" captures the gap. The Amex GBT-SAP Concur platform launched late 2025 with quarterly feature releases planned - promising in ambition but years behind Navan in execution.
03 - Above the Market
Platform Giants Reshaping Travel Distribution
The real threat to incumbents comes not from each other but from the platform layer above. Four technology giants now control the consumer's first moment of travel intent, and each is moving to extend control through the booking and payment layers.
Greatest Existential Risk
Google
16-18% of desktop searches show AI Overviews
93% zero-click rate in AI Mode (Semrush)
Organic CTR drops 61% when AI Overview appears
Travel booking via AI Mode launching 2026
Google poses the greatest existential risk to travel distribution. Travel AI Overview coverage grew 381% after the March 2025 core update. Google's AI Mode is now building agentic flight and hotel booking directly within search, partnering with Booking.com, Expedia, Marriott, and Wyndham. Google insists it has "no intention of becoming an OTA" - but keeping the entire planning-to-booking journey inside Google's ecosystem effectively makes partners into back-end suppliers.
Commerce Stack Builder
OpenAI
800M weekly active ChatGPT users
Booking.com + Expedia as first travel partners
Operator: autonomous browsing agent for bookings
ACP co-developed with Stripe
OpenAI is constructing a complete commerce stack. The Apps SDK launched October 2025 using MCP. Operator navigates any website without API integrations - a "post-API" paradigm that bypasses traditional distribution entirely. OpenAI's Instant Checkout, co-developed with Stripe, enables in-chat purchasing for 1M+ Shopify merchants. Travel checkout infrastructure is in place; activation is a matter of when, not if.
Protocol Layer Control
Anthropic
MCP: foundational connectivity standard
10,000+ active MCP servers globally
~100M SDK downloads per month
Adopted by OpenAI, Google, Microsoft
Anthropic doesn't compete for consumer travel but controls the foundational protocol layer. MCP, open-sourced in November 2024 and donated to the Linux Foundation in December 2025, has become the de-facto standard for agent-tool connectivity. Travel-specific MCP implementations include Kiwi.com, Sabre, Turkish Airlines, Expedia, and Booking.com.
Sleeping Giant
Apple
1.5B daily Siri requests
2B+ active Apple devices
Gemini embedded via Apple Intelligence (Jan 2026)
Controls travel intent at lock-screen moment
Apple is the sleeping giant. A confirmed partnership with Google to embed Gemini into Apple Intelligence (January 2026) positions Google closer to the travel intent moment on every iPhone globally. If Apple structures Siri as a transaction layer - expected with iOS 26 - it could become the dominant mobile travel booking interface overnight. Meta, with 700M+ consumers using Meta AI and 2.7B WhatsApp users, presents massive distribution potential in emerging markets.
04 - Protocol Architecture
Four Protocols Building the New Distribution Stack
The agentic commerce infrastructure is organising into a layered stack. Convergence is occurring at the protocol level but fragmentation persists at implementation. All major protocols use JSON-RPC 2.0. All are open-source under Apache 2.0. But no single protocol spans the full travel commerce stack from discovery to multi-currency settlement.
MCP
Connectivity Layer
Model Context Protocol - How AI agents access external tools and data. Launched by Anthropic, now governed by the Linux Foundation. Solves the NxM integration problem: one MCP server makes inventory accessible to every AI platform simultaneously. 10,000+ servers, ~100M SDK downloads/month.
Anthropic (author)OpenAIGoogleMicrosoft
A2A
Agent Communication
Agent-to-Agent Protocol - How independent agents built on different frameworks collaborate. Enables multi-agent travel systems where a planning agent, pricing agent, booking agent, and support agent coordinate. MCP is how agents call tools; A2A is how agents talk to each other.
Google (April 2025)150+ orgs
ACP
Transaction Layer
Agentic Commerce Protocol - Complete checkout flows inside AI conversations. Three components: Agentic Checkout Spec (5 REST endpoints), Delegated Payment Spec (scoped single-use tokens), Commerce State Spec (persistent shopping state). Critical gap: supports only single merchant per session - insufficient for multi-supplier travel itineraries.
OpenAI + Stripe1M+ Shopify merchants
NDC / ONE Order
Travel Data Layer
IATA Standards - The industry-specific data layer AI agents need to search and book. NDC at 60-65% adoption across airlines. ONE Order (unified PNR/e-ticket records) at less than 27% - mass adoption unlikely before 2030. Investment requirement: $50-100M per carrier over 3-7 years.
NDC: 60-65% airlinesONE Order: <27%
05 - The New Buying Moment
When the Agent Books: How Discovery and Commerce Transform
BCG describes agentic commerce as "the biggest disruption to retail since the birth of online shopping." The fundamental shift is the collapse of the traditional funnel. Search, comparison, and consideration compress into a single agent-mediated moment. The consumer expresses intent; the agent delivers a completed transaction.
Traditional Funnel - 7 Steps
Awareness / Intent
Search and Discovery
Comparison
Consideration
Checkout
Payment
Confirmation
Agentic Funnel - 4 Steps
Intent - Agent receives goal
One-shot research + comparison
Confirmation (human-in-loop)
Autonomous payment + booking
Zero-click commerce is already measurable
Google's AI Mode produces a 93% zero-click rate. 58.5% of US Google searches now end without any click. On mobile the figure reaches 77%. Publisher traffic from Google declined by a third in 2025 - this is not a future risk; it is current-state revenue destruction. Google Flights overtook KAYAK in monthly visitors in under five years, and 80% of Google Flights referrals now go to airline websites rather than OTAs.
Loyalty faces a structural challenge
BCG: "Loyalty becomes less about emotional affinity and more about machine-readable policy." When agents optimise on narrow criteria - price, reviews, delivery speed - brand sentiment becomes secondary. Loyalty programs must become API-first data contracts, not marketing tools. An agent weighing Marriott versus Hilton will parse cancellation flexibility and upgrade probability, not brand affinity.
The new distribution gatekeepers
AI platforms now control which suppliers appear in agent outputs. Being "agent-discoverable" - via MCP servers, structured data, ACP compliance - is the new SEO. Companies without agent-readable supply surfaces are invisible to the fastest-growing distribution channel in travel history.
06 - The Last Mile
Agentic Payments: The Critical Infrastructure That Doesn't Yet Exist
Without the ability to move money, AI agents remain expensive search engines. Payments are the last mile that transforms AI from advisory to autonomous, and the travel industry faces uniquely complex payment challenges that no current protocol fully addresses.
Travel is arguably the hardest vertical for agentic payments. A single trip involves flights, hotels, ground transport, and activities from multiple suppliers across multiple currencies - none of which current ACP was designed to handle.
Four major payment infrastructure initiatives are racing to define this layer. Stripe's Shared Payment Tokens (SPTs) create single-use, scoped tokens constrained by maximum amount, expiry time, and merchant scope. Visa Intelligent Commerce (launched April 2025 with 100+ partners) provides agent-specific tokens with passkey-based authentication enforced at the network level. Mastercard Agent Pay uses registered agent credentials with dynamic token verification - all U.S. Mastercard cardholders were enabled by November 2025. Google's AP2 introduces cryptographic mandates - tamper-proof digital contracts proving user authorization - and supports cards, bank transfers, stablecoins, and crypto.
Critical gaps remain. ACP supports only single merchant per session - a round-trip with two airlines, a hotel, and a car rental is four separate, uncoordinated transactions. Multi-currency settlement across suppliers in different jurisdictions has no standard delegation mechanism. Refund, cancellation, and rebooking authorization chains involve different principals than the original booking authority. Edgar Dunn classifies agent-to-agent transactions in travel as "far from reaching the required level of maturity."
Stripe - Live
Shared Payment Tokens
First implementation of OpenAI's Delegated Payment Spec. Single-use tokens constrained by amount, expiry, currency, and merchant scope. Works across ChatGPT and any ACP-compliant platform.
Visa - Launched
Intelligent Commerce / Agent Pay
Passkey-based authentication, agent identity verification, and transaction controls enforced at network level. Visa has completed hundreds of secure agent-initiated transactions in pilot.
Sabre + PayPal + Mindtrip - Q2 2026
Industry's First End-to-End Agentic Travel
GDS inventory + payment infrastructure + AI planning. First integrated stack from planning intent to confirmed GDS booking with payment. Critical test case for multi-supplier, multi-currency agentic travel.
Amadeus Outpayce - In Development
Agentic Payments Suite
Specifically addressing the multi-supplier settlement challenge unique to travel. Cross-border complexity across suppliers, currencies, and tax treatment is the architectural problem Outpayce is built to solve.
07 - Historical Precedent
What Happens When You Don't Move
Every distribution disruption in travel and adjacent industries follows a recognisable pattern: a new intermediary captures the consumer relationship, incumbents dismiss the threat as niche, and by the time the data is undeniable, adaptation windows are measured in years, not decades. The pattern that should command board-level attention: each disruption wave is getting faster.
Traditional agencies vs. OTAs
1999-2013 - 14 years
59% of storefronts closed
59% of US retail travel agency locations closed (22,938 to 9,387 storefronts). Catalyst: airline commission cuts 2002. OTAs grew from zero to 45% of global bookings. Booking Holdings reached $22B in annual revenue.
Uber vs. taxi medallions
2011-2018 - 7 years
84% value destruction
NYC medallions peaked at $1.05M in 2013-2014. Within five years: $160,000 - an 84% decline. Some sold for $25,000 at trough (97.6% value destruction). Uber and Lyft now control 73% of New York's paid rides.
Google zero-click search
2019-2025 - 6 years
58.5% zero-click searches
58.5% of US Google searches now end without a click to any website. Publishers lost a third of Google traffic in 2025 alone. Zero-click is now the default state of AI-mediated search. Google Flights overtook KAYAK in monthly visitors in under five years.
Agentic commerce - travel
2025-2030 estimated - unfolding now
UNFOLDING NOW
GenAI traffic to travel sites: +4,700% YoY. Google AI Mode: 93% zero-click. 80% of travel executives plan to deploy agentic AI within three years. The clock started in 2025.
"The pattern commanding board attention: each disruption wave is getting faster. Newspapers took 15 years to lose 80% of revenue. Taxi medallions lost 84% in just four to five years. Agentic commerce may move faster still. Incumbents get a two-to-three-year window after the threat becomes visible. After that window closes, the damage is structural and irreversible."
08 - The Blueprint
The Agent-Ready Blueprint: What Travel Companies Must Build Now
A travel company that wants to survive the agentic shift needs five capabilities built in sequence, each enabling the next. This is derived from what leading companies are already implementing and what the emerging multi-rail payment protocol standards require.
01
Agent-Readable Supply SurfacesMonths 0-6Critical - Start Now
AI agents cannot book what they cannot read. Structured, machine-readable product catalogs with real-time pricing exposed via APIs - not human-optimised web pages. Policies (cancellation, baggage, room types) must be machine-parseable. For airlines, this means NDC-compliant offer and order APIs. Sub-200ms response times are the baseline performance requirement. Without this, agents will route demand to competitors whose inventory they can read. This is table stakes - not a competitive advantage.
02
Multi-Platform Tool ConnectivityMonths 3-9High - Enables Distribution
Build MCP servers exposing search, availability, booking, and management tools. This single investment makes inventory accessible to Claude, ChatGPT, Gemini, Copilot, and every other MCP-compatible platform simultaneously. Implement ACP checkout endpoints for OpenAI's Instant Checkout ecosystem. Publish A2A AgentCards for agent-to-agent discovery. Sabre estimates custom integration with each AI agent takes 6+ months without standardised protocols; MCP server deployment can be accomplished in days for simple use cases.
03
Embedded Checkout and Delegated PaymentsMonths 6-12Hardest Technical Challenge
Accept Stripe Shared Payment Tokens and Visa agent-specific tokens. Implement scoped authorization flows with programmatic constraints - maximum amount, currency, merchant, expiry, and revocability. Build payment orchestration that handles multi-supplier travel itineraries within agentic checkout sessions. For companies processing payments across multiple currencies and supplier networks, this requires a payment infrastructure layer capable of routing across cards, open banking, stablecoins, and APMs - optimising for cost, speed, and settlement efficiency per transaction.
04
Trust, Safety and AuditabilityMonths 6-18Regulatory Requirement
Visa's Trusted Agent Protocol and Mastercard's Agent Pay Acceptance Framework establish the baseline: agent identity verification, passkey authentication, transaction controls enforced at network level, and Commerce Signals logging the consumer's original instruction alongside purchase details for dispute resolution. OTAs must build audit trails connecting every agent-initiated action to a verified consumer authorisation. Human-in-the-loop approval flows remain essential for high-value transactions - the 2% autonomous booking rate signals consumers will want oversight for years to come.
05
Commercial Strategy for Agent RoutingMonths 12-24Existential Strategic Choice
The critical question every OTA must answer: do you become the execution layer (an API backend that AI agents call for inventory) or a proprietary agent (a branded AI that owns the consumer relationship), or both? Booking Holdings is pursuing both simultaneously. New monetisation models include agent routing fees, preferred supplier placement for AI recommendations, API-call-based pricing, and outcome-based commissions tied to successful bookings rather than clicks. The strategic imperative is multi-protocol from day one - no single platform will dominate, and hedging across OpenAI, Google, and Anthropic ecosystems is essential.
09 - Conclusion
Conclusion: The Two-Year Window
The agentic travel disruption is not a future scenario to monitor - it is an active restructuring with measurable data points. The full commerce stack - discovery via MCP, coordination via A2A, checkout via ACP, payment via SPTs - exists in production for retail and is being extended to travel now.
01
$3-5 trillion
McKinsey's agentic commerce projection by 2030 = 3-5x the current online travel market
02
80% of executives
plan to deploy agentic AI within three years while only 2% have deployed beyond pilot
03
93% zero-click rate
in Google AI Mode - the default state for AI-mediated search that travel has no strategy to counter
04
4-12 weeks
to go from concept to production with proven agentic payment implementation methodology
05
2-year window
to build infrastructure before agent routing decisions by major AI platforms become structural and fixed
The companies that will define the next decade of travel distribution are making irreversible infrastructure decisions right now. Those that move first - building agent-readable supply, MCP connectivity, agentic payments, and trust infrastructure - will capture disproportionate share of a structurally larger market.
Historical disruption parallels are unambiguous: incumbents get two to three years from the moment the threat becomes visible. That clock started in 2025. Explore Yugo's agentic payments for travel infrastructure and see how the AI payments stack is being assembled in production today.
