Artificial intelligence is moving beyond travel inspiration. Agentic systems can interpret a goal, consult several sources, compare options and use tools to complete a task. In travel, that task may involve building an itinerary, tracking a fare, booking a hotel, changing a flight or arranging an in-trip service.
- AI could become the interface that organises the whole journey, rather than a better search box
- Trust, payments, live availability and support when something fails still constrain autonomous booking
- Travel retail needs to compete for visibility before the passenger reaches the airport
Once an agent can understand the traveller's mission and execute the next step, commercial power shifts towards the system controlling data, recommendation logic and access to the transaction.
What is happening
The threat is now part of the strategic discussion at the largest travel intermediaries. On 20 July 2026, the Financial Times examined Expedia's response to autonomous agents that can plan and book trips. Four days later, Reuters reported the concern in direct commercial terms. Search, comparison and aggregation, which sit at the centre of the online travel agency model, are tasks an AI agent is designed to perform.
The shift is visible in operating products. Booking.com has linked language models to proprietary information on pricing, availability, policies and reviews. Its AI Trip Planner grew from a prototype launched in ten weeks. Omio connects conversational interfaces to live inventory covering rail, coach, ferry and air travel. Expedia Group B2B is developing components that allow other brands to embed travel supply and services inside their own AI workflows.
This does not mean autonomous agents already complete most bookings. Expedia's research with YouGov shows a clear gap between planning and purchase. Travellers are willing to use AI for suggestions, itineraries and price monitoring, yet many still prefer to transact with a familiar brand. The transition is likely to be gradual and hybrid. An agent may research and recommend while an airline, hotel, OTA or payment platform owns checkout, servicing and accountability.
Why it is happening now
Conversation handles complex intent
Traditional search works when the customer knows the destination, dates and product. Many journeys begin with a less precise request, such as a warm short break without crowds, a quiet hotel near public transport or a practical gift that can be collected before departure. Conversation can combine budget, timing, preferences, restrictions and context in one request.
Transaction infrastructure is already available
The language model is only the visible layer. A usable recommendation needs inventory, fares, schedules, policies, identity, payments and service. Expedia Group B2B says its platform handles 21 billion API calls per day and supports 75,000 partners. Infrastructure of this scale allows an agent to move from generic advice to an option that is live, eligible and bookable.
Distribution is spreading beyond travel websites
A booking opportunity can appear inside a general assistant, mobility app, card programme, loyalty environment or airport platform. Each new interface competes to capture intent before the traveller opens an OTA or visits an airline website.
Intent
Traveller mission, limits, timing, preferences and context.
Data
Live price, stock, eligibility, schedules, policies and location.
Decision
Ranking rules, commercial logic, relevance and disclosure.
Action
Booking, payment, collection, fulfilment and confirmation.
Service
Changes, disruption, support, handover and accountability.
Measure
Sales, margin, satisfaction, support demand and incrementality.
The visible assistant is only one layer. Advantage comes from connecting intent, inventory, transaction, service and measurement.
What it means for tourism and travel retail
The first change is about timing. Commercial decisions move earlier. An agent can recommend insurance, transfers, an eSIM, lounge access, baggage, food or travel products while assembling the itinerary. By the time the passenger reaches the airport, some choices may already have been made.
The second change concerns visibility. A physical store gives shelf space to a range of products. A conversational answer may present three. Product attributes, availability and ranking rules become a new form of merchandising. Retail media changes too. A brand may pay for visibility, but the recommendation must remain relevant, disclosed and capable of producing incremental sales rather than paid exposure alone.
The third change is analytical. Conversations reveal intent before the transaction. Questions about weather, baggage, children, connections, health or gifting can provide early demand signals by route and occasion. These signals do not replace bookings, flight schedules or sales. They enrich them and can shorten the time between a market change and a commercial response.
Opportunities by company type
Make products readable by AI
Structure attributes, pack roles, occasions, restrictions, claims, price and availability. Build bundles around destination, trip length, climate and mission.
Compete before the terminal
Connect catalogue, stock, eligibility, collection and promotions with assistants that only recommend fulfilable options.
Become the context layer
Provide terminal location, security time, distance, opening hours, disruption and commercial capacity.
Join travel and commerce
Use booking, check-in, connection and disruption context to present relevant services with consent and clear rules.
Sell beyond the room
Expose policies, amenities, food, wellness, mobility and experiences as structured, combinable services.
Influence discovery
Improve data on seasonality, access, events, neighbourhoods and alternatives so recommendations fit the traveller's intent.
Enter the itinerary
Integrate insurance, eSIM, mobility, finance, tax refund and assistance into a coherent travel proposition.
Build the decision layer
Connect identity, feeds, inventory, rules, attribution, sponsored content and handover to human support.
Risks and barriers
- Incorrect recommendations. An outdated price, stock position, policy or restriction can invalidate the whole experience.
- Opaque influence. Travellers need to distinguish organic relevance from paid promotion and commercial relationships.
- Fragmented accountability. When the agent, platform, supplier and payment provider are different organisations, problem resolution can become unclear.
- Privacy and control. Personalisation improves with trip, loyalty and purchase data, but requires clear purpose, minimisation and consent.
- Platform dependency. Companies may lose direct traffic, customer data and margin if they surrender the interface without a distribution strategy.
- Availability bias. An agent may favour the easiest inventory to integrate rather than the best option for the traveller or destination.
How Marksyte can help
Marksyte can help turn agentic AI into a measurable commercial system rather than an isolated conversational layer.
Demand forecasting
Combine flights, bookings, sales, searches and conversational signals to forecast demand by route, category and time.
Route and destination analysis
Estimate how new services, connections and AI recommendations change the value of each origin and market.
Traveller segmentation
Translate intent, mission, time and context into actionable segments that go beyond demographics.
Assortment, inventory and pricing
Prioritise products, packs, stock and promotions according to expected demand, margin and fulfilment confidence.
Retail media and personalisation
Design visibility rules, separate organic and sponsored recommendations and measure incremental value.
Automation and AI assistants
Build controlled prototypes for planning, recommendation, support and internal analysis with human escalation.
Operational planning
Connect the forecast with staffing, opening hours, replenishment, collection points and disruption response.
Commercial impact measurement
Use experiments, control groups and attribution models to separate convenience, cannibalisation and true growth.
The aim is not to automate every decision. It is to identify where AI removes friction, where it improves precision and where a person or trusted brand should remain in control.
Marksyte can assess data readiness, demand, assortment, distribution and measurement before an agentic solution is scaled.
Agentic AI does not remove the need for brands, retailers or intermediaries. It changes how they prove value. Businesses that provide reliable data, real availability, accountable service and a clear proposition can become part of the new interface. Those that do not may remain present in the market while disappearing from the decision.
Suggested charts and secondary images
From search to execution
Six-step diagram: intent, data, decision, booking, service and measurement.
SEO alt text: Agentic AI travel value chain from traveller intent to commercial measurement
The trust gap
Bars comparing AI use for suggestions, itineraries and price monitoring with willingness to book.
SEO alt text: Gap between using artificial intelligence for travel planning and trusting it to book
A connected traveller itinerary
Editorial illustration of a mobile interface connecting flight, hotel, transfer, airport, duty free and destination.
SEO alt text: Traveller uses an AI assistant to organise flights, hotel and travel retail purchases
The new digital shelf
Visual comparison between a physical shelf with many products and an AI answer with three recommendations.
SEO alt text: Comparison between airport retail assortment and products selected by artificial intelligence
Frequently asked questions
How is agentic AI different from a travel chatbot?
A chatbot answers or recommends. An agentic system can break a goal into steps, consult tools, compare alternatives and take actions such as monitoring prices, booking or changing a service within defined controls.
Will agentic AI replace online travel agencies and airport retail?
There is no evidence for complete replacement. Search and comparison face greater pressure. Businesses that own trusted inventory, service, fulfilment and transaction capability can still capture value even when the interface changes.
What should a travel retail business do first?
Audit product, price, stock, eligibility, location and collection data. Then test one specific traveller mission and connect recommendation quality with conversion, margin and fulfilment.
Sources
- Financial Times, Expedia CEO's defence against AI, 20 July 2026
- Reuters, World Cup boom, Middle East gloom could split earnings for US travel firms, 24 July 2026
- Expedia Group and YouGov, The AI Trust Gap, 14 April 2026
- Expedia Group, Q1 2026 results filed with the U.S. SEC
- Expedia Group B2B, AI toolkit and travel distribution platform, 20 May 2026
- OpenAI and Booking.com, personalising travel at scale
- OpenAI and Omio, the future of conversational travel, 23 June 2026
- OpenAI, Introducing ChatGPT agent, 17 July 2025
- OpenAI, Instant Checkout and the Agentic Commerce Protocol, 29 September 2025
The trust figures come from research commissioned by Expedia Group and conducted by YouGov among more than 5,700 adults in the United States, United Kingdom and India in March 2026. Omio development-speed results are company estimates reported in the OpenAI case study.