---
title: "SI in Travel and Hospitality: 2027 Guide | Webtures"
description: "Explore SI in travel and hospitality through agentic booking, dynamic pricing, guest experience and operations in this 2027 readiness guide from Webtures."
source_url: "https://www.webtures.com/insights/super-intelligence-integration-in-the-travel-and-hospitality-industry/"
lang: "en"
updated: "2026-09-29T00:00:00.000Z"
---

# SI Integration in Travel and Hospitality: A Guide to Preparing for 2027

Short Answer From discovery to booking, pricing to operations: a 2027 readiness guide for travel brands built on accurate data, tiered agent authority and measurement.

Webtures 29 min read

Summarize with SI

![SI Integration in Travel and Hospitality: A Guide to Preparing for 2027](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/cover.svg?v=si1) ![SI Integration in Travel and Hospitality: A Guide to Preparing for 2027](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/cover-light.svg?v=si1)

Webtures Growth & GEO

Published: 20 May 2026 Updated: 29 Sept 2026

From discovery to booking, from operations to the guest experience: global developments, limits of authority and practical readiness steps.

**Current assessment: 29 September 2026.** This guide combines official statistics, regulatory texts and company announcements with Webtures’ strategic assessment. The sections on 2027 are readiness recommendations and scenarios; they should not be read as results that have already happened.

SI integration in the travel and hospitality industry covers far more than adding a chat window to a website. It affects how a destination is discovered, how prices are set, which interface a booking is completed in and how quickly a problem during a trip is resolved.

What the sector sells is not a product waiting on a shelf but capacity that will be used on a specific date and loses its value once that date has passed. An unsold room night or a seat left empty after departure cannot be sold later. That is why a small error in price, availability or cancellation terms affects guest trust, distribution cost and revenue at the same time.

In Webtures’ approach, preparing for 2027 starts with four questions: **Are your brand and properties represented accurately in SI answers? Are the price and availability shown valid at the moment of transaction? When an agent makes a booking, change or refund, what authority is it acting on? Does this experience produce guest satisfaction and sustainable revenue?**

## Executive summary: what should the priorities be when preparing for 2027?

- **Connect price, availability and rules data to a single source of truth.** Room type, rate, cancellation terms, taxes and additional fees must not contradict one another on the brand website, in the channel manager or in SI interfaces.

- **Accept that bookings will start in more than one interface.** The brand website, online travel agencies (OTAs), metasearch and SI assistants will continue to coexist.

- **Give agents tiered authority, action by action.** Providing information, preparing options, creating a booking, making a change, initiating a refund and taking payment are different levels of risk.

- **Manage dynamic pricing within the limits of transparency.** A price that changes with demand and a price set on the basis of personal data are not the same practice; the total price and the terms must be clear.

- **Build the 2027 budget on measurable pilots that fit the seasonal cycle.** Test in a quieter period and decide whether to expand or stop before the peak season.

## What is SI integration in the travel and hospitality industry?

SI integration is the connection of forecasting, recommendation, content generation or task execution capabilities to a business’s real data and processes. A model writing a fluent travel itinerary is not the same capability as accessing current availability in a hotel’s reservation system or creating a booking securely.

| Approach | What it means in travel and hospitality | Measure of success |
| --- | --- | --- |
| Predictive SI | Likelihood of demand, cancellations and no-shows; maintenance and staffing needs | Forecast error and the outcome of the decision based on the forecast |
| Recommendation systems | Recommending destinations, room types, add-on services and packages | Incremental contribution after cancellations and guest satisfaction |
| Generative SI | Multilingual guest replies, content drafts, review summaries | Accuracy, response time and human correction rate |
| Agentic SI | Using tools for search, booking, change and refund steps | Share of tasks completed correctly and with authority |

These approaches can work together: a demand model forecasts occupancy, the team makes the pricing decision and an authorised agent processes a date change. Not every automation is SI either; a channel manager distributing a price to every channel is a rule-based operation and needs integration testing.

## What does the 2026 data say?

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

Official statistics on travel demand and company announcements on SI use measure different things. The table below summarises the size of the sector and its direction in 2026; no row shows the impact of SI on its own.

| Indicator | Verified information | How should it be read? |
| --- | --- | --- |
| International tourist arrivals, 2025 | UN Tourism estimated around 1.52 billion international tourists in 2025, an increase of 4% | An estimate published in January 2026; it may change with later revisions |
| International tourism receipts, 2025 | UN Tourism’s preliminary estimate is USD 1.9 trillion, an increase of 5% | A preliminary estimate; total tourism export revenue including passenger transport is given separately as USD 2.2 trillion |
| First half of 2026 | UN Tourism estimated around 690 million international tourists in January–June 2026 and reported a 22% drop in the Middle East | The year is not complete; the annual growth expectation was lowered from 3–4% to 1–2% |
| Economic weight of the sector, 2025 | According to WTTC, travel and tourism contributed USD 11.6 trillion to global GDP, a 9.8% share, supporting 366 million jobs | The total of direct, indirect and induced impacts; a model estimate that should not be added to UN Tourism data |
| Aviation, 2026 | IATA forecast 5.1 billion passengers, USD 23.0 billion in net profit and a 2.0% net margin for 2026 | A June 2026 forecast, lowered from the previous expectation because of disruption in the Middle East and high fuel prices |

[UN Tourism’s January 2026 release](https://www.untourism.int/news/international-tourist-arrivals-up-4-in-2025-reflecting-strong-travel-demand-around-the-world) describes strong demand in 2025, while [its September 2026 update](https://www.untourism.int/news/international-tourism-holds-steady-with-04-growth-in-first-half-of-2026-as-middle-east-conflict-and-rising-costs-weigh-on-momentum) shows that growth almost stalled in the first half. [WTTC’s April 2026 release](https://wttc.org/news/travel-tourism-sees-best-year-ever,-outpacing-the-global-economy-in-2025) calculates the economic contribution with its own impact model, and [IATA’s June 2026 outlook](https://www.iata.org/en/pressroom/2026-releases/06-07-middle-east-disruptions-high-fuel-prices-halve-airline-industry-profitability/) is a forecast.

From Webtures’ perspective, the common message is this: as demand growth slows, margin pressure rises; SI investment should be justified by its measurable effect on distribution cost, conversion and service capacity.

## How do SI assistants and agentic booking work in travel planning?

![The booking journey: discovery, comparison, verification, confirmation, payment and after-sales, each with a verification question; price and availability are re-verified at the moment of transaction (representative flow)](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-1.svg?v=si1)![The booking journey: discovery, comparison, verification, confirmation, payment and after-sales, each with a verification question; price and availability are re-verified at the moment of transaction (representative flow)](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-1-light.svg?v=si1)

Three models are currently developing side by side in travel planning; in each, the customer relationship and responsibility are distributed differently.

**The brand’s own assistant:** According to the company, [Hilton AI Planner](https://stories.hilton.com/releases/hilton-introduces-the-hilton-ai-planner) became available to all hilton.com visitors on desktop and tablet from 17 March 2026. [Marriott’s Ask Bonvoy experience](https://www.prnewswire.com/news-releases/marriott-international-introduces-ask-bonvoy-a-new-ai-powered-search-experience-transforming-travel-exploration-302801647.html) launched in June 2026 as a beta, in US English and for a subset of members. [Booking Holdings’ second-quarter 2026 remarks](https://s25.q4cdn.com/383369491/files/doc_events/2026/08/Q2-26-Prepared-Remarks-for-Website.pdf) report that an SI-powered discovery experience is being tested on Booking.com and that the next generation of Priceline’s agentic assistant, Penny, is being deployed.

**A brand app inside a general SI assistant:** According to [OpenAI’s October 2025 announcement](https://openai.com/index/introducing-apps-in-chatgpt/), Booking.com and Expedia were among the first pilot partners for apps in ChatGPT; at first, apps were available to logged-in users outside the European Economic Area, Switzerland and the United Kingdom. [In May 2026, Expedia](https://www.expedia.com/newsroom/plan-your-next-trip-with-expedia-in-claude/) offered a similar connection for Claude in the US; users see options with live prices in the chat and complete the booking on Expedia. According to [IHG’s 2026 half-year results](https://www.ihgplc.com/~/media/Files/I/Ihg-Plc/results/2026/half-year-results-2026/stock-exchange-announcement.pdf), the company’s ChatGPT app also directs guests to IHG’s direct channels.

**Agentic booking on an SI search surface:** [On 27 August 2026, Google](https://blog.google/products-and-platforms/products/search/book-travel-ai-mode/) announced that hotel booking in AI Mode had started rolling out gradually in the US, in English. Payment is made with Google Pay and the hotel or booking platform remains the merchant of record; the first partners include Booking.com, Expedia, Hilton, IHG, Marriott and Trip.com. For flights, Google offers prices and price tracking, while tickets are bought from the airline or a preferred platform. IHG states that bookings from this channel are processed as direct bookings for the company.

User trust is not developing at the same pace. In [Expedia Group’s April 2026 survey](https://ir.expediagroup.com/news-and-events/news/news-details/2026/Expedia-Group-Reveals-The-AI-Trust-Gap-Travelers-Embrace-AI-for-Planning-but-Rely-on-Trusted-Brands-to-Book/default.aspx) of more than 5,700 adults in the US, the UK and India, 66% said they would not trust an SI assistant to buy or book anything on their behalf; only 8% felt comfortable booking through an SI platform. It is a company survey limited to three countries; even so, it shows that getting help with planning and granting booking authority are different decisions.

The difficulty of agentic booking is that a request contains many conditions. For a guest who says “Find a hotel near the airport with free cancellation for two adults and a child in mid-October”, the system must verify the following:

- The dates, the number of guests and whether the room suits that number of guests.

- The total price including taxes and mandatory fees, and the currency.

- The cancellation deadline and when payment will be taken.

- Availability confirmed from a live source at the moment of transaction.

- After confirmation, the booking number and the route for changes and cancellation are communicated.

For flights, connection times, baggage allowance and matching the passenger name to the identity document are added. These checks relate more to the transaction design of the booking system than to the language ability of the model. For the equivalent flow in retail, see the guide to [agentic commerce in retail and e-commerce](https://www.webtures.com/insights/si-integration-in-retail-and-e-commerce/); the difference in travel is that capacity is tied to a date and prices can change within minutes.

## Revenue management and dynamic pricing: where should the line be drawn?

Airlines and hotels started using revenue management systems long before SI was on the agenda. SI is not reinventing this field; it aims to improve demand forecasting by assessing more signals together, such as event calendars, search trends and cancellation behaviour.

Two practices must be distinguished. A **dynamic price** changes for everyone according to demand, occupancy and timing for the same date and product. A **personalised price** means showing a different price for the same product based on a person’s profile, history or device. The latter can create additional disclosure obligations and a risk of discrimination.

The Delta example shows why this distinction matters. On its [July 2025 earnings call](https://s2.q4cdn.com/181345880/files/doc_earnings/2025/q2/transcript/CORRECTED-TRANSCRIPT_-Delta-Air-Lines-Inc-DAL-US-Q2-2025-Earnings-Call.pdf), Delta’s management said that the SI-assisted pricing it runs with Fetcherr was still in the test phase, stood at about 3% of domestic and had a year-end target of about 20%. In its [August 2025 statement](https://news.delta.com/delta-responds-misinformation-around-ai-pricing), the company said it does not use, test or plan any fare product that sets individual prices for customers on the basis of personal data. These are company statements; whether the target was reached has not been verified in this guide. The lesson is clear: pricing that cannot be explained clearly to customers and regulators creates a trust problem.

A control framework for SI-assisted pricing should include:

- Price floors and ceilings, brand policy and rules for price consistency across channels.

- Competitor price data coming from a source that can be used legally and contractually.

- Automatic price increases being stopped in exceptional situations such as natural disasters or transport disruption.

- A record of the model’s recommendation, the price applied and who approved it.

- A clear total price, including taxes and mandatory fees, from the first display.

In the first phase, model recommendations with human approval can be used; as they are proven through price complaints, cancellation rates and contribution after commission, businesses can move to automatic updates within defined limits.

## Guest experience and communication

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

In guest communication, the goal is to solve the problem correctly and at a sustainable cost; a conversation ending without being passed to a human agent does not prove that the problem was solved. Answering questions before booking carries relatively low risk; flight disruption, refunds and complaints are sensitive moments, both emotionally and financially.

Company announcements show the scale. According to [Airbnb’s second-quarter 2026 shareholder letter](https://s26.q4cdn.com/656283129/files/doc_financials/2026/q2/Airbnb-Q2-2026-Shareholder-Letter.pdf), almost 45% of issues that start with its SI assistant are resolved without a human agent, and the assistant works in more than 50 languages; support-related cost per booking fell by around 16% year on year, which the company attributes in part to improvements in the assistant. [Booking Holdings reported in the same quarter](https://s25.q4cdn.com/383369491/files/doc_events/2026/08/Q2-26-Prepared-Remarks-for-Website.pdf) that it had extended voice SI support to most eligible incoming calls and that customer service cost per booking is still falling at a double-digit rate. These are company statements; one brand’s rate should not be used as another business’s budget assumption.

In multilingual communication, the translation of statements such as cancellation terms, allergen information or safety instructions should be tested separately. At handover to a human, a summary of the conversation should be passed on; human support should always be available for special assistance requests and guests in vulnerable situations. If complaints rise while support costs fall, the application should be reassessed.

## Hotel operations: housekeeping, maintenance and energy

In operations, the value of SI appears in processes the guest does not see but notices immediately: the room being ready on time, a fault being fixed before a complaint and energy being managed without compromising comfort.

**Housekeeping:** Cleaning order can be prioritised by assessing occupancy, check-out times, early check-in requests and room types together. The critical distinction is whether planning is based on rooms or on people. Under the EU AI Act, systems that allocate tasks based on individual behaviour or personal traits, or that monitor and evaluate employee performance, are classified as high-risk under Annex III; the application date for this category is 2 December 2027. Planning based on rooms and workload is a lower-risk starting point.

**Maintenance:** Sensor data from air conditioning, lifts and water heating systems can show early signs of faults. The value of a prediction depends on it being linked to the right work order and spare part; a system that produces many false alarms loses the trust of the technical team.

**Energy:** Heating, cooling and lighting can be managed together with occupancy, the booking calendar and the weather. Measurement should be made per occupied room night and together with guest comfort complaints. A saving achieved by the guest finding the room too cold or too hot is not a real gain.

In all three areas, the first piece of work is often matching data in the PMS, the building management system and the maintenance software to the same room and time identifiers.

## Verified examples from aviation operations

A large share of SI claims in aviation comes from supplier presentations or from news with unclear sources. The examples below are the companies’ own statements; they have not been independently audited, and one airline’s result cannot be transferred to another.

**Route planning:** According to [Alaska Airlines’ August 2024 announcement](https://news.alaskaair.com/sustainability/how-ai-is-helping-alaska-airlines-plan-better-flight-routes-and-lower-emissions/), the tool called Flyways identified route improvement opportunities on 55% of flights and delivered 3–5% fuel savings on flights of more than four hours. Its recommendations are presented to dispatchers and pilots as decision support. The announcement is two years old and should not be read as a current result.

**Predictive maintenance:** [In April 2025, Lufthansa Technik](https://www.lufthansa-technik.com/en/latam-opts-for-lufthansa-technik-s-digital-platform-aviatar-cecbda242cc7c47c) announced that predictive analytics on its AVIATAR platform were being used on more than 300 LATAM aircraft and that first results showed 20% fewer delays and cancellations. As no measurement period is given, it should be read as an early company finding.

**Employee tools and crew planning:** The five-year partnership [Ryanair announced with Google Cloud in August 2026](https://corporate.ryanair.com/news/ryanair-google-cloud-announce-five-year-data-and-ai-partnership/) aims to give 35,000 employees access to generative SI tools and to optimise flight crew logistics. This is an announcement; it contains no outcome data.

In these examples, SI does not make the operational decision on its own; it offers the expert team a recommendation or an early warning. On the passenger side, the most sensitive moment is disruption: a rebooking suggestion can be produced quickly, but compensation and exception decisions should remain tied to clear rules and authorised staff.

## Loyalty programmes and personalisation

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

Loyalty programme data is among the most valuable first-party data travel brands hold. Past stays, room preferences and communication consents can be used for more accurate offers and a lower-friction experience.

The limit of personalisation is the guest’s expectation: carrying business travel habits straight into a family holiday can turn into the wrong recommendation. Recommendation success should be measured alongside additional sales against cancellations and subsequent stay behaviour, with a control group where possible.

A personalised offer and a personalised price should also be kept apart. A member-only benefit is a clearly defined programme rule; changing the price of the same room from person to person based on profile data requires separate disclosure and risk assessment.

## Content and image generation: where is the line with reality?

Generative SI can save time in translating property descriptions, drafting content and summarising reviews. The line is clear: guests expect the room and the view they see to be the same in reality; room size, distance to the beach and accessibility information must come from a verified source.

In the EU, [Article 50 of the AI Act](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50) requires businesses deploying a system that generates or manipulates image, audio or video content resembling existing persons, objects, places or events, and that would falsely appear authentic, to disclose that the content is artificial. Presenting a view that a property does not have as a genuine photo carries a risk of misleading consumers in addition to this obligation.

Review summaries should not create praise or complaints that do not appear in real reviews, nor hide negative themes. Fake guest reviews or synthetic testimonials presented as real experience should not be part of this system. A good content workflow uses approved property data, flags missing information and keeps a record of the published version.

## How should levels of authority be defined for agentic SI?

![Levels of authority for agentic SI: providing information is read-only, options and offers are recommendations, booking and payment are approved actions, changes are rule-based, cancellation and refunds are policy-based, compensation and exceptions are human decisions](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-2.svg?v=si1)![Levels of authority for agentic SI: providing information is read-only, options and offers are recommendations, booking and payment are approved actions, changes are rule-based, cancellation and refunds are policy-based, compensation and exceptions are human decisions](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-2-light.svg?v=si1)

Agentic SI is an SI system carrying out multi-step tasks by using tools within the goal and authority set by the user. In travel, these tasks quickly reach actions that are hard to reverse and have financial consequences. Authority should therefore be defined action by action, not through the question “can the agent make bookings?”.

| Action | Recommended starting level | Approval requirement | Core control |
| --- | --- | --- | --- |
| Providing information (availability, rules, amenities) | Read-only | None | Verification from a live source and the date of the information |
| Preparing options and offers | Recommendation | None | Total price, taxes, fees and cancellation terms shown together |
| Creating a booking and taking payment | Approved action | Explicit customer approval and strong authentication | Confirmation of amount and currency; preventing duplicates on repeated requests |
| Changes (dates, room, passenger details) | Rule-based action | Re-approval if there is a fare difference or penalty | Fare rules, name change restrictions, inventory check |
| Cancellation and refunds | Policy-based action | Human decision for cases outside the policy | Calculating the refund amount, fraud checks, record-keeping |
| Compensation and exception decisions | Human decision | Authorised staff | Relevant passenger and consumer rights, documents and rationale |

The table is implementation guidance, not a legal classification. A repeated request should not create two bookings; if the price changes between the offer and payment, the system should not quietly proceed at the old amount.

Standards reflect this distinction too. Under the [Universal Commerce Protocol (UCP)](https://ucp.dev/documentation/announcements/), a lodging technical council made up of Amadeus, Booking.com, Expedia, Google, Hilton, Marriott and Trip.com was formed in August 2026. [UCP’s lodging booking specification](https://ucp.dev/draft/specification/lodging/booking/) is still a draft and provides that, unless the AP2 extension that carries payment authority in a verifiable way is supported, the booking must be completed manually by the user in a trusted interface. On the card network side, [Mastercard and Trip.com presented in September 2026](https://www.mastercard.com/news/eemea/en/newsroom/press-releases/en/2026/september/mastercard-and-trip-com-pave-the-way-for-agentic-commerce-in-the-travel-sector) a solution offering end-to-end search, booking and payment through the TripGenie assistant; the commercial launch is expected in early 2027, so it is not yet in general use.

To separate which problem each protocol solves, the [agentic commerce protocol map](https://www.webtures.com/insights/agentic-commerce-protocols-how-mcp-a2a-acp-ucp-and-ap2-work-together/) is a good starting point. Connecting to a protocol does not make price data correct, and it does not guarantee payment acceptance or more bookings.

## Data protection and payment security

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

Travel data is more sensitive than an ordinary e-commerce order; passport details, travel dates and accommodation address show where and when a person will be. Some requests can also become special category data: accessibility and special assistance requests may be health data, and religiously motivated meal preferences may reveal beliefs. [Article 9 of the GDPR](https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng) lists health data, data revealing religious beliefs and biometric data processed to uniquely identify a person among the special categories. This information should be processed only as far as the service requires and with limited access, and should not be passed into the context of a general-purpose model more than necessary.

For applications such as biometric boarding and automated check-in, the AI Act excludes biometric verification limited to confirming that a person is who they claim to be from the high-risk biometrics category in Annex III. This exception does not change the special status of biometric data under data protection law; the legal basis, an alternative process and the retention period must be designed separately.

A guest message or a review may contain text that tries to change the agent’s task; content from outside sources should not be treated as a command, and authority to act should be limited by server-side rules.

On the payment side, card data should never reach the model; tokenisation, payment fields hosted by the payment provider and per-transaction authorisation are the core principles. For businesses that process, store or transmit card data, the future-dated requirements of [PCI DSS v4.x](https://blog.pcisecuritystandards.org/now-is-the-time-for-organizations-to-adopt-the-future-dated-requirements-of-pci-dss-v4-x) have been mandatory since 31 March 2025.

## Technical architecture: how do the PMS, CRS, GDS, channel manager and NDC connect?

![Technical architecture: PMS, CRS, channel manager, GDS, NDC and revenue management, CRM and payment systems connect to the SI assistant and agent through a control layer covering access, business rules, approval and logging (representative diagram)](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-3.svg?v=si1)![Technical architecture: PMS, CRS, channel manager, GDS, NDC and revenue management, CRM and payment systems connect to the SI assistant and agent through a control layer covering access, business rules, approval and logging (representative diagram)](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-3-light.svg?v=si1)

In travel, the success of an SI project is often decided in the connection layer before the model; the price of the same room reaches the guest through more than one system.

- **PMS (property management system):** Runs on-property operations; room status, the guest record and the folio are held here.

- **CRS (central reservation system):** The central source of rates, availability and reservations.

- **Channel manager:** Distributes price and availability to OTAs and other sales channels.

- **GDS (global distribution system):** Distribution networks through which travel agencies access flight, hotel and car inventory.

- **NDC:** [As defined by IATA](https://www.iata.org/en/iata-repository/pressroom/fact-sheets/fact-sheet-ndc/), an API-based data communication standard through which sellers shop, order, pay and service with airlines using the Offer and Order standards. [In August 2026, Lufthansa Group](https://lhgroupairlines.com/ndc/en/about-ndc/news-room) said that more than half of its sales through travel agencies were made via NDC.

- **Revenue management, CRM and payment systems:** The pricing decision, the guest profile and collection run in these layers.

| Information | Source of truth | Freshness requirement |
| --- | --- | --- |
| Room price and availability | CRS or channel manager | At the moment of transaction |
| Flight offer and ancillary services | The airline’s offer system (NDC or GDS) | Within the offer’s validity period |
| Cancellation, change and refund rules | Fare rules | Per fare |
| Property amenities and accessibility | Approved property content | At every change |
| Booking and order status | PMS, CRS or airline order system | At the moment of transaction |

A [retrieval-augmented generation (RAG) approach](https://www.webtures.com/insights/retrieval-augmented-generation/) can be used for property rules and frequently asked questions; however, it does not replace live price and availability checks. Current transaction information should be obtained by connecting to the relevant system.

Distribution technology companies are also opening this layer to agents. According to [Sabre’s September 2026 statement](https://investors.sabre.com/news-releases/news-release-details/mcp-new-ndc-travels-shift-streaming-moment), the MCP server it launched in September 2025 allows SI agents to book, sell and service travel on Sabre’s data and inventory, with nearly 80 customers in pilot or full production. This is a company statement. An MCP server does not define an agent’s authority by itself; access, business rules, approval and logging layers must be designed by the business.

## Regulation: which dates matter when preparing for 2027?

There is no single “SI compliance date” for travel businesses operating globally; obligations vary with the system’s intended purpose, the business’s role and the market. The summary below is not legal advice.

The provisions of the EU [AI Act](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689) on prohibited practices have applied since 2 February 2025; systems that infer emotions in the workplace are among these prohibitions, except for medical or safety reasons. With the regulation’s general date of application, 2 August 2026, the transparency obligations in Article 50 also took effect. [Amending Regulation 2026/1744](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202601744) then changed the timetable item by item:

- For high-risk systems under Annex III, the application date is 2 December 2027.

- For high-risk systems linked to regulated products under Annex I, the date is 2 August 2028.

- Providers of systems that generate synthetic content and were placed on the market before 2 August 2026 must comply with the marking obligation in Article 50(2) by 2 December 2026.

In travel, these dates translate into SI disclosure for assistants that talk to guests, disclosure for generated images that could be taken as real and high-risk obligations for systems that allocate tasks to employees.

In EU consumer law, informing consumers that a price was personalised on the basis of automated decision-making is an obligation added to Directive 2011/83 by [Directive 2019/2161](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32019L2161). Because the directive applies only in part to passenger transport and package travel, its scope should be assessed separately for hotel, flight and package products. Package travel rules have also been revised: [Directive 2026/1024](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202601024) provides for the new rules to apply from 29 March 2029. The European Commission’s Digital Fairness Act initiative had not yet been published as a proposal as of 29 September 2026.

In the US, the Federal Trade Commission’s (FTC) [rule on fees](https://www.ftc.gov/news-events/news/press-releases/2025/05/ftc-rule-unfair-or-deceptive-fees-take-effect-may-12-2025) took effect on 12 May 2025 and requires hotels and short-term lodging to show a total price that includes mandatory fees.

## How is success measured?

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

The measurement system should combine technology performance with the guest experience and the commercial result. An application that improves one indicator can create a hidden cost in another area.

| Dimension | Core indicators | Balancing measure to track alongside |
| --- | --- | --- |
| Discovery | Visibility in SI answers, accuracy of property and price information | Rate of incorrect amenity, price or rules information |
| Booking | Conversion rate, share of direct bookings, net room revenue | Cancellation rate and contribution after commission |
| Revenue management | Revenue per available room (RevPAR), unit revenue in airlines | Price complaints, fare inconsistency, long-term demand |
| Guest experience | First-contact resolution, response time, satisfaction | Repeat contacts and the quality of handover to humans |
| Operations | Room readiness time, intervention before failure, energy per occupied room night | Employee workload and guest comfort |
| Reliability | Share of actions completed correctly and with authority | Duplicate bookings, incorrect refunds and human intervention |

Denominators must be clearly defined: is conversion calculated per session, and within which time window are cancellations deducted? Because of seasonality, comparisons should be made with the same period, similar properties or routes and, where possible, a control group.

## How is return on investment calculated? A hypothetical example

**Hypothetical calculation.** The calculation below is not a Webtures client result or an industry average. It is a hypothetical scenario prepared to show how an investment model can be built.

Assume that on the website of a hotel group receiving 200,000 eligible visits a month, an SI-assisted search and question-answering feature raised booking conversion from 1.50% to 1.65% in a controlled test. The increase is 0.15 percentage points, or 10% in relative terms.

| Calculation item | Assumption | Result |
| --- | --- | --- |
| Additional bookings | 200,000 × (1.65% − 1.50%) | 300 bookings/month |
| Net contribution per booking | USD 60 after cancellations, payment costs and variable expenses | USD 18,000/month additional contribution |
| Ongoing SI operating cost | Total of model, monitoring and operations | USD 6,000/month |
| Net monthly benefit | 18,000 − 6,000 | USD 12,000 |
| Initial investment | Integration and preparation | USD 84,000 |
| Simple payback period | 84,000 / 12,000 | 7 months |

The calculation assumes a constant monthly volume; for seasonal properties it should be rebuilt with the annual traffic distribution, and the implementation period and cost of capital should be added. Some of the additional bookings might have come through OTAs or the call centre anyway; attributing the whole difference to the SI investment is not correct until channel shift is separated from genuine additional demand.

## The first 90 days: how should a start that fits the seasonal cycle be made?

![A 90-day starting plan: scope and baseline on days 1–15, data and authority limits on days 16–30, a limited pilot on days 31–60, impact and decision on days 61–90; test in a quieter period and decide before the peak season](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-4.svg?v=si1)![A 90-day starting plan: scope and baseline on days 1–15, data and authority limits on days 16–30, a limited pilot on days 31–60, impact and decision on days 61–90; test in a quieter period and decide before the peak season](https://www.webtures.com/images/insights/artificial-intelligence-integration-in-the-travel-and-hospitality-industry/section-4-light.svg?v=si1)

The first phase should focus on producing a measurable result in a single use case rather than on a broad tool purchasing programme. In travel, timing is part of the decision: starting a pilot in the middle of the peak season both distracts teams and mixes the result with seasonal effects.

| Period | Work to be done | Exit criterion |
| --- | --- | --- |
| Days 1–15 | Use case, current performance, data sources and owners are defined | A measurable problem and a baseline |
| Days 16–30 | Price, availability and rules data are audited; authority limits and risk scenarios are written | Verified data and written authority limits |
| Days 31–60 | A pilot runs with a limited set of properties, routes or languages | A predefined quality and reliability threshold |
| Days 61–90 | Commercial impact is assessed against a control group; a decision is made before the peak season | A decision to expand, correct or stop |

Stopping criteria should be written from the outset: the application is restricted when incorrect price or cancellation information, unauthorised actions, duplicate bookings or a negative contribution are seen.

## 2027 roadmap: what should be done in which quarter?

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

The 2027 plan should start from the verified current situation and should not depend on future features that platforms have announced. The timeline is written for businesses whose peak season is summer in the northern hemisphere; those with a different seasonal cycle should shift the same sequence to their own calendar.

**Final quarter of 2026, strengthening the foundations:** Consistency of price, availability and property content is audited; an inventory of current SI uses is taken; a baseline is measured for SI visibility; data use and exit options in supplier contracts are reviewed.

**First quarter of 2027, producing evidence:** Selected pilots are tested in a controlled way during a quieter period; commercial impact, guest experience and the cost of errors are assessed together.

**Second quarter of 2027, connecting channels:** Suitable SI booking surfaces, direct connections and guest communication are integrated; load and outage scenarios are tested before the peak season.

**Third quarter of 2027, controlled operation in the peak season:** Monitoring is strengthened instead of adding new scope; human support capacity and handover flows are maintained.

**Final quarter of 2027, portfolio decision:** Profitable applications are scaled, and those that do not produce results are closed or redesigned. For applications that may fall within high-risk scope in the EU, the timetable and scope are re-verified with the legal team; the 2028 budget is linked to verified value.

## SI visibility for travel brands: is your information represented accurately?

A hotel’s visibility in an SI answer breaks down into levels: is it named, is it recommended, is it cited as a source, is the information correct and can the user move on to booking?

In travel, accuracy problems often stem from outdated information: a changed check-in time or pet policy can live on in an old page or an OTA listing and reach the guest through an SI answer.

A monitoring system records the query, date, platform, country, language and answer. The query set should reflect real travel intent: “hotel in Rome” and “family-friendly hotel near the airport with late check-out” are different commercial opportunities. Recommended indicators include:

- The visibility rate of the brand and its properties across the relevant query set.

- The accuracy rate of information on amenities, rules, location and accessibility.

- How current price and availability statements are, and whether they match the price at the moment of transaction.

- The distribution of cited pages and visibility compared with competitors.

- The booking and contribution outcome of trackable SI referrals.

On some platforms, price accuracy is a direct condition of visibility. [Google’s hotel price accuracy policy](https://support.google.com/hotelprices/answer/6064419) requires the total price on the booking page to match the total price shown on Google and to include mandatory taxes and fees; a low accuracy score can reduce visibility, and serious violations can lead to property or account suspension. The same discipline is also the core measure of how reliable prices in SI answers are.

Accessibility information needs particular care; incorrectly conveyed access information directly affects a guest’s trip. It should be published from a single source, measured and dated.

Webtures builds this measurement through its [Visibility Intelligence](https://www.webtures.com/visibility-intelligence/) approach, based on the query set, accuracy audits and correction priorities. The aim is not simply to increase visibility but to make sure guests reach the right information at the moment of decision.

## Frequently asked questions

### Where should travel businesses start with SI?

Start with a measurable problem with clear boundaries. Repeated guest questions, inconsistencies in property content or housekeeping delays can be suitable starting points. A solution should not be chosen before the current cost of the problem and the quality of the data are known.

### How can you tell whether a hotel website is ready for agentic booking?

Check whether price, availability, room type, cancellation terms, taxes and additional fees are presented consistently and in a form machines can read. The integrations the booking engine supports and whether repeated requests create duplicate bookings should also be tested. For a comprehensive framework, see the [Agentic Commerce Readiness](https://www.webtures.com/agentic-commerce-readiness/) assessment.

### Do SI assistants show hotel and flight prices accurately?

The price shown depends on the source the assistant connects to and how current that source is. An offer retrieved through a live connection and a price summarised from an old page are not equally reliable. The final amount, including taxes and fees, should be confirmed again at the booking step.

### Can an SI agent book and pay on a customer’s behalf?

It may be possible on supported platforms and under certain conditions; the country, payment method, the seller’s integration and the authority granted by the customer are decisive. The scope of authority and spending limit should be clear, and strong customer approval should be maintained at the payment step.

### Does dynamic pricing mean a personalised price?

No. A price that changes for everyone with demand, dates and occupancy is a different practice from a price set according to a person’s profile. A personalised price can create additional disclosure obligations and a risk of discrimination in the target market.

### Does SI reduce dependence on OTAs?

Not by itself. SI assistants draw on both brand websites and OTAs. The share of the direct channel can be protected through accurate information, a competitive total price, a smooth booking experience and strong loyalty value.

### Will front desk and call centre teams no longer be needed in 2027?

No firm conclusion can be drawn in that direction. Repeated questions and standard transactions can be automated; disruption, complaints, special needs and exception decisions will continue to need human support.

## The Webtures approach: from visibility to booking

[GEO](https://www.webtures.com/generative-engine-optimization-geo/) · SI VisibilityIs your brand visible in generative search?Let's build a strategy to surface your brand in ChatGPT, Gemini and Perplexity answers.[Get in touch →](https://www.webtures.com/contact/)

Free AssessmentMeet a digital strategy team operating since 2011Share your goals and we'll map a visibility roadmap tailored to your brand.[Get in touch →](https://www.webtures.com/contact/)

Measurable GrowthUnite GEO and performance marketing in one modelLet's generate sustainable digital demand with a data-driven approach.[Get in touch →](https://www.webtures.com/contact/)

The approach Webtures recommends is to assess SI visibility together with accurate information, the guest journey, agentic commerce readiness and measurement. A travel brand being discovered, understood correctly and booked reliably are parts of the same commercial process.

The first step is to establish the current situation: how are you represented in SI answers, where do your price and property details conflict, and what obstacles are there in the booking journey? The second step is to set priorities, and the third is to move forward by measuring results. Technical implementation should be handled together with the business’s technology teams and distribution providers; responsibilities for payments, data protection and legal compliance should be clearly defined with the relevant parties.

**Build your 2027 readiness on data-driven priorities.** To assess your SI visibility, the accuracy of your information and the opportunities in your booking journey, get in touch with Webtures.

[Get in touch →](https://www.webtures.com/contact/)

Webtures Growth & GEO

Published: 20 May 2026 Updated: 29 Sept 2026

[Add Webtures as a preferred source on Google](https://www.google.com/preferences/source?q=webtures.com)

## Articles related to SI Integration in Travel and Hospitality: A Guide to Preparing for 2027

[### Agentic Commerce Readiness Report: how commerce infrastructure changes Webtures / 22 Sept 2026](https://www.webtures.com/insights/agentic-commerce-readiness-report/)

[### How to Build a GEO Strategy for Cosmetics and Beauty Brands? Tufan Acar / 09 Sept 2026](https://www.webtures.com/insights/how-to-build-a-geo-strategy-for-cosmetics-and-beauty-brands/)

[### Agentic Commerce Protocols: How MCP, A2A, ACP, UCP and AP2 Work Together Atiye Berika Ertaş / 28 Aug 2026](https://www.webtures.com/insights/agentic-commerce-protocols-how-mcp-a2a-acp-ucp-and-ap2-work-together/)

[### What Is Dark SI Traffic? The Measurable and Hidden Side of SI-Driven Traffic Atiye Berika Ertaş / 28 Aug 2026](https://www.webtures.com/insights/what-is-dark-si-traffic-the-measurable-and-hidden-side-of-si-driven-traffic/)

[### Content Optimization for SI-Powered Search Engines Atiye Berika Ertaş / 15 Aug 2026](https://www.webtures.com/insights/content-optimization-for-si-search/)

[### What Is Zero-Shot Learning? When Does It Provide an Advantage? Atiye Berika Ertaş / 15 Aug 2026](https://www.webtures.com/insights/what-is-zero-shot-learning/)