EU AI Act transparency duties apply now; high-risk duties from December 2027. Check your exposure
Insights About us Careers
Contact us
Industries / 8 verticals

AI for Travel and Hospitality

The sector that invented dynamic pricing, now being asked to explain the difference between pricing the seat and pricing the passenger. Eight verticals, written for the person who owns the revenue system rather than the one presenting the innovation roadmap.

8 verticals
Individual pages
11 Aug 2026
Eight airlines questioned
Seat
Not the passenger
Why this sector is different

Dynamic pricing and personalised pricing are not the same thing, and the distinction has stopped being academic.

Travel has priced by demand, timing and inventory for decades, and that is ordinary revenue management with a long and well-understood history. Pricing by who the individual buyer is — their device, their browsing history, their loyalty profile, data purchased about them — is a different proposition, and it is now under direct scrutiny. On 11 August 2026 a House committee sent detailed information requests to eight US airlines covering data collection for pricing, third-party data purchases, use of machine learning to individualise prices, disclosure and opt-out mechanisms, and segmentation by income, race, age and behaviour. That expanded an inquiry opened on 11 May 2026 covering twenty-five major corporations.

The practical consequence for anyone building a revenue system here is that the question is no longer whether the model improves yield. It is what the model uses, whether that can be explained, and whether the answer survives being read aloud. We have told operators that a feature improving revenue slightly is not worth the sentence they would have to write about it in a disclosure, and that is a commercial judgement as much as a legal one — but it has to be made deliberately rather than discovered when the letter arrives.

Evidence, not enthusiasm

Where AI actually earns its place in travel and hospitality

Ranked by evidence rather than by how often it appears in a conference keynote. The maturity column is our own read; the catch column is what the vendor deck leaves out.

WhereWhat it doesMaturityThe catch
Demand forecasting and revenue managementPredicts booking demand and sets inventory and price by segment.ProvenThe most mature commercial application anywhere in this sector, with decades of method behind it. The inputs, not the technique, are now the question.
Service automation and enquiry handlingHandles booking, amendment and information requests.ProvenHigh volume and genuinely effective. Disruption is where it fails, and disruption is when customers most need it.
Translation and multilingual serviceServes guests and travellers in their own language.ProvenAmong the highest-value applications in a sector whose customers are by definition foreign.
Operational disruption managementRebuilds schedules, crew and guest arrangements after failure.StrongSubstantial value and genuinely hard. Constraints are legal and contractual, not preferences.
Personalisation of content and offersMatches product, content and offers to a traveller.StrongEffective for relevance. Personalising the price rather than the offer is where it becomes a different question.
Review and feedback analysisExtracts themes and issues from unstructured guest feedback.ProvenReliable and underused. Most operators read reviews and never analyse them at property or route level.
Housekeeping and workforce schedulingMatches labour to forecast occupancy and service demand.StrongReal savings. Working time rules and schedule stability belong in the constraints, not in the trade-offs.
Maintenance and asset conditionMonitors aircraft, vessel, vehicle and property condition.GoodWorks as condition monitoring against known-good. Failure prediction runs into the same missing-failures problem as everywhere.
Individualised pricing from personal dataSets a price for a specific person using data about them.Under scrutinyDisclosure obligations in force in some jurisdictions and active investigation in others. Distinct from demand-based pricing.
Automated compensation and claims decisionsDecides passenger or guest entitlement without review.ConstrainedPassenger rights reform adds notification and response duties that assume a process, not just an outcome.

The disruption paradox is the sector's real service problem

Automated service handles routine volume well and degrades exactly when it matters most. A cancelled flight, a missed connection, an overbooked property or a medical situation aboard generates enormous contact volume, high emotion, complex entitlements and non-standard resolutions all at once — and a system tuned on routine enquiries meets that traffic at its weakest. Operators consistently measure containment in normal operations and discover the failure during an event, when the customer relationship is actually decided. The design answer is to build for the disruption case first, route aggressively to people when entitlement or distress is involved, and treat normal-operations containment as the easy half rather than the achievement.

What the rules require

Pricing scrutiny, passenger rights and accommodation data

This sector answers to consumer pricing and disclosure regulation, passenger rights regimes, accommodation registration and data rules, and gambling regulation where relevant. This is our reading as at September 2026 and we work alongside your legal, commercial and compliance functions rather than in place of them.

Surveillance pricing scrutiny

Eight airlines questioned in August 2026

On 11 August 2026 a House committee sent nineteen detailed questions to eight US airlines covering customer data collection for pricing, third-party data purchases, use of AI and machine learning to individualise prices, disclosure and opt-out mechanisms, loyalty programme data affecting pricing, and segmentation by income, race, age and behaviour. It expanded an inquiry opened on 11 May 2026 into twenty-five major corporations, following an FTC surveillance pricing study published in January 2025 which found that consumer characteristics and behaviours could be tracked and used to tailor pricing.

Algorithmic pricing disclosure

Already law in New York

New York's Algorithmic Pricing Disclosure Act has required, since 10 November 2025, that a price set by an algorithm using a consumer's personal data be displayed with a specific disclosure statement, with penalties up to $1,000 per violation. Several other states have introduced or enacted comparable measures. The obligation attaches to personal-data-driven pricing rather than to demand-based pricing, which makes the distinction between the two a compliance question and not a semantic one.

Air passenger rights reform

Adopted July 2026, applying around August 2027

Political agreement was reached on 15 June 2026 and the reform formally adopted by Council on 13 July 2026, applying roughly twelve months and twenty days after publication. Compensation is unchanged — the three-hour threshold and €250, €400 and €600 bands remain. What is new is process: rerouting must be offered within three hours at no cost in comparable conditions or the passenger may self-arrange and claim up to 400% of the ticket price; airlines must notify eligible passengers electronically of their rights, claim guidance and the disruption reason within 96 hours of arrival; and claims must be answered within 30 days by paying or justifying refusal.

Short-term rental data

Applying since 20 May 2026

Regulation (EU) 2024/1028 has applied since 20 May 2026. Hosts register online and receive a unique registration number; platforms must display and verify registration numbers, carry out random checks, and share monthly data on guest stays and nights booked with authorities through a Single Digital Entry Point in each Member State, with authorities able to require removal of non-compliant listings. It operates on an opt-in basis — Member States need not apply it unless they introduce registration or request platform data, at which point they must do so through this Regulation.

Consumer protection in booking flows

Design is regulated, not just price

Drip pricing, pressure messaging, misleading scarcity claims and dark patterns in booking journeys are enforcement priorities in several jurisdictions, and the new passenger rights rules require hand baggage prices to be displayed upfront before booking. A conversion optimisation programme that learns to apply urgency pressure is optimising directly into an enforcement risk, which is worth stating before the experiment framework is built.

Gambling harm regulation

Where casino and gaming operations are involved

Gambling regulators increasingly expect operators to identify and act on markers of harm, and the same behavioural modelling capability can be pointed at harm detection or at maximising play. Which way it points is an operator decision with regulatory, ethical and licensing consequences, and the two objectives are not compatible in one system.

What this means for a build

Three consequences. Revenue systems need a clear, documented boundary between demand-based and person-based inputs, because that line is what a disclosure obligation or an information request turns on. Service automation should be designed for disruption first and routine second, since disruption is when entitlement decisions and customer relationships are actually made. And anything deciding a passenger or guest entitlement now sits inside a process with notification and response deadlines attached, which makes the workflow the deliverable rather than the decision. See AI policy development and AI risk assessment.

Questions

FAQ

Marked up with FAQPage schema so these answers can surface in search results and inside AI assistant responses.

Do you have travel and hospitality experience?

Yes in demand forecasting and revenue analytics, service automation and multilingual handling, review and feedback analysis, and operational scheduling. Less in airline network planning specifically, and none in reservation system replacement — we would not claim otherwise. Each of the eight vertical pages states our depth in that area rather than implying uniform expertise across a sector spanning a national carrier and a tourism board.

Is dynamic pricing still safe to use?

Demand-based pricing is ordinary revenue management with decades of precedent, and nothing has changed about it. What has changed is scrutiny of pricing driven by data about the individual buyer — device, browsing history, purchased third-party data, loyalty profile. New York has required a specific disclosure on personal-data-driven pricing since 10 November 2025, and in August 2026 eight US airlines received detailed questions about exactly this. The practical step is to document which inputs your system uses and which side of that line each falls on, because that is the question you will be asked.

Where does service automation actually fail?

During disruption, which is when it matters most. Routine enquiries are high volume and easy, and containment rates in normal operations look excellent. A cancellation, a missed connection or an overbooking produces a surge of emotional, complex, entitlement-laden contacts that a system tuned on routine traffic handles badly. We would design for the disruption case first and route aggressively to people wherever entitlement or distress is involved, treating normal-operations containment as the easy half rather than as the result.

What does the passenger rights reform change for our systems?

Process rather than compensation. The three-hour threshold and the €250, €400 and €600 bands are unchanged. What is new, from around August 2027, is that rerouting must be offered within three hours or the passenger may self-arrange and claim up to 400% of the ticket price; that eligible passengers must be notified electronically of their rights, claim guidance and the disruption reason within 96 hours of arrival; and that claims must be answered within 30 days. Those are workflow and data obligations, and the 96-hour duty in particular requires knowing who was eligible without waiting for them to ask.

What should a travel business build first?

Review and feedback analysis, in most cases, because it is the cheapest route to knowing what is actually wrong. Operators read reviews and almost never analyse them systematically by property, route, cabin, season or issue type, which means recurring operational problems stay anecdotal. It touches no pricing question, requires no new data collection, and it routinely identifies fixable issues that were costing more than any model would save. Multilingual handling is a close second in a sector whose customers are by definition foreign.

Start with the problem, not the technology.

Thirty minutes, no charge, no deck. Tell us what is going wrong in your organisation and we will tell you whether AI is the right instrument — including when it plainly is not.