Industries
AI in Hospitality & Tourism 2026: 12 Use Cases
AI in hospitality 2026: 12 use cases for hotels, holiday rentals and destinations with benefit, risk and effort, plus AI Act, GDPR and a 90-day plan.

Key takeaways
In short:
- AI in hospitality in 2026 mostly means answering guests faster, producing copy and translations faster and taking pressure off the front desk. Hospitality on site remains a human job.
- Only around 26% of European hotels use AI so far, and small independent properties are falling behind (HOTREC 2026).
- The biggest levers for a 30–80-room hotel: multilingual guest messaging, AI phone reception, pre-arrival upselling and visibility in AI answers.
- Since 2 August 2026, chatbots and voice AI must disclose that they are AI (Article 50 AI Act). Allergies and accessibility needs are usually health data under Article 9 GDPR.
- A realistic start takes 90 days: pilot two use cases, measure, then expand.
AI in hospitality and tourism is the use of language models, image AI and automated workflows along the guest journey: from inspiration through enquiry and booking to the stay, the review and the next booking. According to Eurostat, EU accommodation businesses that used AI in 2025 used it mostly for marketing and sales (58.82%). Hotels themselves are more hesitant: in the HOTREC European Hotel Distribution Study 2026, around 26% of hotels across Europe use AI, 59% do not yet and 15% plan to.
The reason is telling. According to HOTREC, the gap is organisational rather than technical: many hotels rate their IT as solid but lack AI budgets, structured data, staff training and clear responsibilities (HOTREC 2026). That is where this guide comes in. It is vendor-neutral and written for hotels, guesthouses, holiday rental operators, tour operators and destination organisations across Europe. How to save commission and win guests directly is covered in our guide to increasing hotel direct bookings; here we focus on the AI tools behind it.
This article is not legal advice. Have specific deployments, especially involving health data, call recording and marketing, reviewed by a lawyer in your country.
Where does AI really help in tourism? 12 use cases at a glance
The table lists the most relevant use cases. "Effort" means the implementation effort for a 30–80-room property: low (feature in your existing PMS or a standard tool), medium (setup, templates, one integration), high (integrating several systems).
| # | Use case | Benefit | Main risk | Effort | Tool category |
|---|---|---|---|---|---|
| 1 | Multilingual guest messaging and chatbot (website, WhatsApp, email) | answers around the clock, fewer guests drifting to OTAs | invented prices or promises, missing AI notice | medium | chatbot, messaging platform, PMS guest messaging |
| 2 | AI phone reception | no missed calls, relief at peak times | recording without notice, AI disclosure | medium | voicebot with PMS and calendar integration |
| 3 | Pre-arrival upselling | extra revenue from parking, breakfast, spa, late check-out | marketing without consent, overselling | low to medium | guest CRM, pre-arrival tool |
| 4 | Review response drafting | faster, more personal replies | generic text, disclosing guest data | low | reputation tool, text AI |
| 5 | Revenue management support | rate suggestions based on demand and competition | black-box prices, reputational damage | medium | revenue management system (RMS) |
| 6 | Content and image production | regular content for website and social media | misleading or unlabelled AI images | low | text and image AI, scheduling tool |
| 7 | Listing and guest info translation | consistent quality in every guest language | wrong facts in one language, legal texts | low | translation AI, channel manager |
| 8 | Housekeeping and maintenance scheduling | rooms ready sooner, less idle time | wrong priorities, data quality | medium | housekeeping app, PMS module |
| 9 | Staff knowledge assistant | faster onboarding, consistent answers | outdated documents, access rights | medium | internal AI assistant with document search |
| 10 | GEO: being recommended by ChatGPT and Google AI Mode | visibility in AI answers | contradictions between channels | medium | website, structured data, profiles |
| 11 | Preparing for agentic trip planning | stay bookable when assistants book | unclear availability in Europe | medium | booking engine, Google Hotel Center |
| 12 | Reporting and KPIs | weekly report instead of spreadsheet work | wrong numbers without sources | low to medium | BI tool, automation |
1. Multilingual guest messaging: the biggest lever
A guest assistant is an AI system that answers questions about arrival, parking, pets, breakfast times, accessibility or cancellation terms in the guest's language, based on your own property information. The benefit is twofold: guests get an instant answer, and reception no longer answers the same thirty questions every day. Before booking, it stops prospects from switching to a platform, as we show in our guide to direct bookings.
Three rules make the difference. The assistant only answers what is in a maintained knowledge base. Prices and availability come live from the booking system or not at all. Special requests, groups and complaints go to a person. If you use WhatsApp, Meta has charged per delivered message since 1 October 2026; prices depend on the recipient's country (for German numbers, a service message costs €0.0456), and the first 1,000 service messages per number and month are free (Meta rate cards). Details are in our WhatsApp Business API 2026 guide.
2. AI phone reception: answering when the desk is busy
Phone and email are still major booking channels. In HOTREC's German sample, 18.1% of 2025 overnight stays came in by phone and 17.8% by email (HOTREC 2026). An AI phone assistant answers calls, handles standard questions, takes reservation requests and hands over to reception or schedules a call-back. Two duties always apply: the AI notice at the start (Article 50(1) AI Act) and no secret recording. National law varies; in Germany, for example, unauthorised recording of private spoken words is a criminal offence under Section 201 StGB. How we build such assistants is described on our AI chatbots & phone assistants page.
3. Pre-arrival upselling: relevant, not loud
AI helps send the right extra to the right guest: parking for guests arriving by car, the spa package for a couple on a weekend break, breakfast for the business traveller on a room-only rate. A message with arrival information is service; pure marketing by email or messenger generally requires prior consent under the EU ePrivacy Directive, with a narrow exception for existing customers (Directive 2002/58/EC, Article 13). Keep service content and offers clearly separate and let the AI personalise only within those rules.
4. Review responses: AI drafts, your tone
Text AI drafts replies to Google, Booking.com and Tripadvisor reviews in seconds, also in the guest's language. A person checks it and adds one concrete detail so it does not read like a template. Never mention room numbers, dates of stay or health details in public replies. Asking for reviews is fine; fake reviews are banned across the EU: submitting or commissioning false consumer reviews is a blacklisted unfair practice under the Unfair Commercial Practices Directive, Annex I No. 23c, whether a human or an AI writes them.
5. Revenue management: check suggestions, do not follow blindly
A revenue management system suggests rates based on on-the-books data, demand, events and competitor prices. For a 30–80-room property, that can save a lot of spreadsheet work. What matters is understanding why a price is suggested and setting floors and ceilings. We do not recommend a specific product here; compare PMS integration, transparency and contract terms. Prices personalised per guest are a special case: for distance contracts, EU consumer law requires you to inform consumers when a price has been personalised on the basis of automated decision-making (Consumer Rights Directive, Article 6(1)(ea)). Have that reviewed before you use it.
6. Content and image production: labelling and honesty
AI drafts room descriptions, blog posts, newsletters and social posts. Be careful with images: under the AI Act, AI-generated or heavily manipulated imagery that resembles real places and would falsely appear authentic is a deepfake (Article 3(60)) and must be disclosed (Article 50(4)). A generated pool that looks better than the real one is also misleading about main characteristics under the Unfair Commercial Practices Directive; a label does not cure that. Show rooms, views and facilities with real photos and use AI images for mood, illustration or clearly recognisable visualisations. Wording examples are in our guide to AI Act Article 50 labelling.
7. Translating listings and guest information
Guests from the Netherlands, Poland or Italy book more readily when descriptions, house rules and arrival information are in their language. Machine translation is mature enough for this. The risk lies in facts: a mistranslated cancellation deadline or pet policy leads to disputes. Keep one source version as the single truth, translate automatically from it and have legal texts such as terms and cancellation policies checked by professionals. Ordinary listing copy is not text on matters of public interest, so it generally does not fall under the text-labelling rule in Article 50(4).
8. Housekeeping and maintenance
From departures, arrivals, room status and late check-out requests, a system can suggest the cleaning sequence and turn defect reports from guest messages or photos into tickets automatically. The benefit depends entirely on data quality in your PMS. Start with defect capture; it is simpler than fully automated staff scheduling.
9. Staff knowledge assistant
Seasonal and new staff ask the same things: where is the spare key, how do I cancel a group booking, which allergens are in tonight's menu? An internal knowledge assistant searches manuals, checklists and process descriptions and answers with a source reference. Up-to-date documents and access rights matter so that not everyone sees everything. Since 2 February 2025, the AI literacy obligation applies to everyone using AI at work (Article 4 AI Act).
10. GEO: how does a hotel get recommended by ChatGPT and Google AI Mode?
Generative engine optimisation (GEO) makes AI assistants name your property when someone asks: "Quiet family hotel by a lake with an indoor pool, dogs welcome". Each assistant sources local data differently. For generative answers Google uses, among other things, Business Profile and Maps data and recommends keeping Business Profile information up to date. According to OpenAI's help centre, ChatGPT uses web search and third-party providers that OpenAI only partly names. Perplexity connected Tripadvisor for hotel information in 2025, according to The Verge. Google says its AI Mode has more than one billion monthly users (Google).
What helps in practice:
- A complete Google Business Profile: hotel attributes (parking, Wi-Fi, pets, accessibility), current photos, replies to reviews.
- Structured data using the schema.org type
Hotel, a subtype ofLodgingBusiness, with properties such ascheckinTime,checkoutTime,amenityFeature,petsAllowed,numberOfRoomsandavailableLanguage. Markup must match the visible text; Google says its AI features require no special schema. - Answer-ready pages: one page per audience or occasion (families, cyclists, meetings, accessibility) with verifiable facts instead of marketing language.
- Consistency: keep name, address, facilities and rules identical on your website, Google, Booking.com, Tripadvisor and the destination's tourism portal.
- Allow the crawlers: sites that block OAI-SearchBot do not appear in ChatGPT search answers, according to OpenAI.
Further reading: local AI search optimisation and Google AI Mode and AI Overviews.
11. Agentic trip planning: what is confirmed and what is not?
Agentic trip planning means an AI assistant does more than suggest: it checks availability and takes the guest through to booking. As of November 2026, this is what is confirmed:
- Google announced at I/O on 19 May 2026 that it is expanding agentic booking in Search to local experiences and services. Search brings together the latest pricing and availability, with direct links to finish booking through the provider of your choice. The rollout was announced for everyone in the US (Google).
- OpenAI introduced apps in ChatGPT in October 2025, with Booking.com and Expedia among the pilot partners. At launch, apps were available to users outside the EEA, Switzerland and the UK; OpenAI said it expected to bring them to EU users "soon" (OpenAI). According to OpenAI's help centre, availability depends on the app, plan and region.
What this means for a European hotel: whether and when assistants will book directly with independent hotels in Europe is open. What is certain is that assistants depend on machine-readable prices and availability. The best preparation is therefore unglamorous: a booking engine synchronised in real time with your channel manager and PMS, and a connection to Google via free booking links in Hotel Center. If you are bookable via Google today, you are ready for the next step.
12. Reporting: weekly report instead of spreadsheet work
Automations pull occupancy, average daily rate, channel mix, reviews and enquiry volume from PMS, channel manager and profiles into a weekly report. AI can summarise anomalies in plain sentences. The numbers themselves must come from the systems, not from the language model; every statement in the report should point to a source.
Which rules apply to AI in hotels? An overview
| Rule | What it means for AI in tourism | Applies since / from |
|---|---|---|
| Article 4 AI Act (AI literacy) | train staff in using AI and document it | 2 Feb 2025 |
| Article 50 AI Act (transparency) | AI notice for chatbots and voice AI; disclosure of deepfakes such as realistic AI images of your property | 2 Aug 2026 |
| GDPR Article 28 | data processing agreement with every AI, chatbot and telephony vendor | since 2018 |
| GDPR Article 9 | allergies, intolerances, disabilities and mobility needs are usually health data: only where needed, generally with explicit consent, stored separately, deleted after the stay | since 2018 |
| ePrivacy Directive, Article 13 | marketing by email or messenger only with consent or within the narrow existing-customer exception | national transpositions |
| Unfair Commercial Practices Directive | no misleading AI images or copy; no fake reviews | national transpositions |
| European Accessibility Act | online booking for consumers is an e-commerce service and must be accessible, including chat widgets in it; microenterprises providing services (fewer than 10 staff and no more than €2 million turnover or balance sheet) are exempt | 28 Jun 2025 |
Breaches of Article 50 can be fined up to €15 million or 3% of worldwide annual turnover, whichever is higher; for SMEs the lower amount applies (Article 99). Call recording rules differ by member state; check national law before you record or transcribe calls.
Health data: the underestimated issue
Guests tell the chatbot without being asked: "My daughter has a nut allergy" or "I use a wheelchair, is the bathroom step-free?". Such information generally falls under the special categories in Article 9 GDPR. Practical rules:
- Answer the question without collecting more. "Is the bathroom step-free?" does not require a diagnosis.
- Use it only for the stay. Pass notes to kitchen or housekeeping, but not into the marketing CRM and never for segmentation.
- Consent and deletion. Only record such information deliberately with explicit consent and delete it after departure.
- Check your vendors. Find out whether chat logs are used to train third-party models and exclude it contractually.
The 90-day plan for a 30–80-room hotel
Days 1 to 30: baseline and selection
- Measure the baseline: for two weeks, log calls, missed calls, emails, chat enquiries, response times and front-desk workload.
- Choose two use cases, typically the guest assistant (1) and review replies (4) or AI phone reception (2).
- Build the knowledge base: the 30–50 most common guest questions with verified answers, house rules, arrival, cancellation terms.
- AI inventory and a short AI policy: which tools, which data, who approves? Sign data processing agreements.
- Train the team (Article 4 AI Act) and document it.
Days 31 to 60: pilot
- Launch the assistant on your website and, if relevant, WhatsApp, with an AI notice and handover to reception.
- Set rules for health data in the bot (do not probe, forward, delete).
- Review replies as drafts only, approved by a named person.
- Complete your Google Business Profile, add
Hotelschema and activate free booking links through your booking engine provider.
Days 61 to 90: measure and expand
- Compare KPIs with the baseline (table below).
- Spot-check 30 bot answers and 10 review replies for errors.
- Ask ChatGPT, Gemini and Google AI Mode ten to twenty test questions about your property and log the results.
- Choose the next use case, for example pre-arrival upselling or translation.
Go-live checklist
- The AI notice appears in the first message or greeting.
- Prices and availability come from the booking system, not the model.
- A handover to a person is possible in every channel.
- No recording without prior notice.
- Health information never ends up in marketing.
- Realistic AI images are labelled; rooms and facilities are shown with real photos.
- The chat widget in the booking flow works with keyboard and screen reader.
How do you measure success? KPIs for AI in hotels
| KPI | Definition | Use case |
|---|---|---|
| Automation rate | enquiries resolved without human intervention ÷ all enquiries | 1, 2 |
| First response time | time to first substantive answer | 1, 2 |
| Missed calls | unanswered calls ÷ all calls | 2 |
| Upsell revenue per arrival | pre-arrival extra revenue ÷ arrivals | 3 |
| Review response rate | replied ÷ all new reviews, plus average response time | 4 |
| Error rate | faulty AI answers in the sample ÷ answers checked | all |
| Direct share | direct room nights ÷ all room nights | 1, 10, 11 |
| Mentions in AI answers | share of test questions in which your property is named | 10 |
| Front-desk time saved | hours per week versus baseline | 1, 2, 9, 12 |
Set targets from your baseline, not from vendor promises. Reliable independent benchmarks for automation rates in hospitality are still scarce.
Common mistakes
- A chatbot without a knowledge base: a bot that invents prices or services does more harm than good.
- No route to a human: guests with complaints or special requests want a person.
- AI images nicer than reality: leads to disappointment, bad reviews and legal risk.
- Health data in the marketing CRM: allergies and disabilities do not belong in segments.
- Too many tools at once: two well-measured use cases beat six half-finished ones.
- Waiting for agentic booking: if you are not bookable via Google today, AI assistants will not book you tomorrow either.
- No baseline: without before figures, you cannot prove anything after 90 days.
Conclusion: start with two use cases where guests wait for answers
AI pays off in hospitality wherever guests are waiting for an answer: on the phone, in chat, after a review and in AI search. The rules in 2026 are clear across the EU: an AI notice in conversations, disclosure of realistic AI images, careful handling of health data and accessible booking flows. HOTREC's data shows that small independent properties risk falling behind. Two well-chosen use cases and clean measurement can change that within 90 days.
Our marketing automation for tourism connects campaigns, booking enquiries and guest communication; we add chatbots and voice AI through our AI chatbots & phone assistants service. A good starting point is the 90-minute AI initial consultation for €490 (incl. VAT, pricing). Book a no-obligation call.
Sources
All sources accessed 6 November 2026.
- HOTREC / HES-SO Valais-Wallis, European Hotel Distribution Study 2026 (full study, chapter "AI adoption and AI readiness"): https://www.hotrec.eu/media/static/files/import/all_news_2026_2026_35/full-study_european-hotel-distribution-study-2026_hotrec.pdf
- Eurostat, Use of artificial intelligence in enterprises (2025 data): https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Use_of_artificial_intelligence_in_enterprises
- AI Act, Articles 3, 4, 50, 99: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50 (other articles at the same address)
- GDPR (Regulation (EU) 2016/679), Articles 9 and 28: https://eur-lex.europa.eu/eli/reg/2016/679/oj
- ePrivacy Directive (2002/58/EC), Article 13: https://eur-lex.europa.eu/eli/dir/2002/58/oj
- Unfair Commercial Practices Directive (2005/29/EC), incl. Annex I No. 23c: https://eur-lex.europa.eu/eli/dir/2005/29/oj
- Consumer Rights Directive (2011/83/EU), Article 6(1)(ea): https://eur-lex.europa.eu/eli/dir/2011/83/oj
- European Accessibility Act (Directive (EU) 2019/882): https://eur-lex.europa.eu/eli/dir/2019/882/oj
- German Criminal Code, Section 201: https://www.gesetze-im-internet.de/stgb/__201.html
- Meta, WhatsApp Business Platform pricing (rate cards from 1 October 2026): https://developers.facebook.com/docs/whatsapp/pricing
- Google, "Google Search's I/O 2026 updates" (19 May 2026): https://blog.google/products-and-platforms/products/search/search-io-2026/
- Google Search Central, AI features and your website: https://developers.google.com/search/docs/appearance/ai-features
- Google Hotel Center Help, Free booking links: https://support.google.com/hotelprices/answer/10472393?hl=en
- schema.org, Hotel and LodgingBusiness: https://schema.org/Hotel, https://schema.org/LodgingBusiness
- OpenAI, "Introducing apps in ChatGPT and the new Apps SDK" (6 October 2025): https://openai.com/index/introducing-apps-in-chatgpt/
- OpenAI Help Center, Apps in ChatGPT: https://help.openai.com/en/articles/12503483-apps-in-chatgpt-and-the-apps-sdk
- OpenAI Help Center, Searching the web with ChatGPT: https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt
- OpenAI, Overview of OpenAI crawlers: https://developers.openai.com/api/docs/bots
- The Verge, "Perplexity partners with Tripadvisor to source hotel info" (9 January 2025): https://www.theverge.com/2025/1/9/24339020/perplexity-tripadvisor-integration-hotels
Frequently asked questions
How many hotels use AI in 2026?
According to HOTREC's European Hotel Distribution Study 2026, around 26% of the European hotels surveyed use AI, 59% do not yet and 15% plan to start. Larger properties, chains, city hotels and 4- and 5-star hotels lead. Small independent hotels with fewer than 30 rooms are underrepresented among adopters.
Does a hotel chatbot have to say it is an AI?
Yes, unless it is obvious to guests. Since 2 August 2026, Article 50(1) of the EU AI Act requires that people are told when they are interacting with an AI system. This also applies to voice assistants on the phone. In practice, one sentence in the first message or greeting is enough, combined with an easy handover to reception.
May an AI assistant store guests' allergies or wheelchair needs?
Only with particular care. Information about allergies, intolerances or disabilities is usually health data and therefore a special category under Article 9 GDPR. Collect it only where needed for the stay, generally with explicit consent, keep it out of your marketing CRM and delete it after check-out.
How does a hotel get recommended by ChatGPT or Google AI Mode?
AI assistants rely on sources they can find and understand: Google on Business Profile, Maps and indexed websites, ChatGPT on web search and data partners among others. A complete Google Business Profile, consistent details across all portals, Hotel structured data and pages with concrete facts for each audience all help. Nobody can guarantee mentions.
Where should a small hotel start with AI?
With two use cases that are quick to measure: a multilingual assistant for the most common guest questions and AI drafts for review replies. Measure enquiries, response times and front-desk workload for two weeks beforehand. After 90 days, let the numbers decide which use case comes next.


