I'm Publishing the AI Tourism Playbook. (What's Inside, and What I Chose to Leave Out)

Over the last eighteen months I've talked with destinations, hotel groups, tour operators and attractions all trying to figure out what to do with AI. One question kept coming back, always in the same form: "Where do I start?"

It's not a trivial question. Neither is the answer.

For years technology companies have been selling tools without stopping to ask whether there was anything underneath ready to receive them. The Phocuswright numbers (Travel Innovation and Technology Trends 2026) spell it out clearly: only 6% of hotel chains have a defined AI strategy at the company level. The average reliability score that hotels themselves give their own AI systems is 4.7 out of 10 (h2c AI Study, 2025). Translated: everyone's buying, few are actually using, almost nobody trusts what they've installed.

On the other side, travellers aren't waiting. 42% of global travellers used AI to plan an itinerary in 2025 (Simon-Kucher, Global Travel Trends 2026). Under 35, the figure tops 50% and grows every quarter. ChatGPT has 800 million weekly active users (Phocuswright, 2026). Google AI Mode went from 100 million monthly users in July 2025 to 75 million daily users in the third quarter of the same year.

The gap is right here, in this scissors spread. The technology is already in travellers' pockets. The organisations that are supposed to serve them are still debating whether it's worth it.

The AI Tourism Playbook exists to close that gap — at least for those who genuinely want to close it.

What it is and what it isn't

The playbook is an online book. Not a PDF. Not a course. Not an upsell for consulting.

Eleven sections, one on Mondays, from 27 April to 15 June 2026. Free to read, one-time registration to access sections B through K. Section A is public, readable with no sign-up required. No marketing sequences, no reselling of email addresses, no promotions dressed up as content.

I wrote it in English deliberately. The potential audience is European and therefore multilingual, but English remains the lingua franca of tourism-tech, and writing in English forces a precision that Italian would walk around more loosely. For those who'd rather read in Italian, articles like this one on Officina Turistica will pick up the main themes week by week.

The first book came out in 2024 on Amazon: Data-Driven Destinations. And when it came out, some chapters were already obsolete. The technology didn't stop and won't stop. A web book can be updated; a printed book can't. This time around I can at least fix a chapter without reprinting.

This playbook is the natural continuation of Data-Driven Destinations. It's what I've learned since then, working with people trying to apply AI in tourism without getting burned. Without a data culture, serious AI adoption is very hard. Without structured data, AI improvises. And when AI improvises in tourism, it's not the vendor who pays the price. You pay it, at the front desk, when a guest complains about something you never promised.

The editorial plan, with commentary

The eleven sections are designed as a sequence, not as interchangeable chapters. You can read them out of order, but they're built to walk a specific path: understand where you actually are, understand where you should be, decide what to do on Monday morning.

Section A · The Foundation · 27 April 2026

What has genuinely changed about how travellers search, decide, book. Five years ago they searched. Today they ask. Tomorrow they'll delegate. The section doesn't deal in hype: it deals with the new operational terrain, the four digital channels a tourism operator works in today, and why the European regulatory framework is a global competitive advantage rather than a brake. This is the public section — readable right now, no registration needed.

Section B · The State of AI in Tourism · 27 April 2026

The state of the art today. Not the one declared in vendor reports, but the real one — with the three structural barriers and the one excuse that no longer holds. Phocuswright records that 1% of travel companies are experimenting with or scaling agentic AI. 23% plan to adopt it within three to five years. The distance between those two numbers is where the real work lives. The section explains why travellers are moving much faster than the organisations that should be serving them, and why this asynchrony is the main risk for whoever stands still.

Section C · The AI Maturity Model · 27 April 2026

A maturity model anyone can self-complete in fifteen minutes. Three honest stages, in order of depth: Exploring, Applying, Orchestrating. Most organisations declare themselves in adoption but are actually in exploration… Some announce "orchestration" while running a chatbot hooked to a CMS that doesn't talk to the PMS. Understanding where you actually are is the prerequisite for not wasting the next eighteen months on initiatives with no foundations.

Section D · Use Cases by Segment · 4 May 2026

Same operational terrain, four different costumes. The technology doesn't change by segment. The operational reality surrounding it does. A DMO doesn't control the product: it coordinates discoverability. A hotel has data density and fragmented systems. A tour operator has specificity as a structural advantage that AI can amplify. An attraction is rich in content and poor in digital channels. For each, three use cases that actually pay off — and what to put off. One example from the hotel block: the top-5 OTA marketing spend was $17.8 billion in 2024 (Phocuswright). Spending on direct AEO channels in the same year cost those who did it two orders of magnitude less and produced measurable results.

Section E · Multilingual GEO and AEO · 11 May 2026

How, in practice, to be visible and recommendable by AI systems. When a traveller asks ChatGPT "where should I sleep in Cefalù," what comes back isn't magic. It's the result of how you've written your site's pages, how consistently your hotel's details match across Google Business Profile and Booking, and how many times you're cited in sources the AI considers reliable. All things you can work on.

The section draws on serious research, conducted by Princeton and IIT Delhi, that tested dozens of techniques across ten thousand real queries in twenty-five different sectors to understand what actually shifts visibility on AI engines. One result I'll preview here: the most effective technique raises visibility by 41%. For small operators — those who in Google tests start at the bottom of the results page — a different technique brings the gain up to 115%. Translated: in AI systems, a well-structured small hotel can appear above a poorly structured large OTA. In traditional search engines this is nearly impossible, and accordingly it doesn't happen.

For European tourism, made up mostly of small operators competing against enormous platforms, this is one of the few genuinely good pieces of news in this transition. The section also includes the part that usually goes missing: how to handle multiple languages at once. A Sicilian hotel needs to be found in English, German and French on the same day, and AI models aren't equally capable in every language. Which to focus on first, and where a human hand is actually needed to reach an acceptable result.

Section F · The Tool Landscape · 18 May 2026

An honest, non-exhaustive map of useful tools for tourism operators. Not a sponsored review. European tools where they exist, global alternatives where needed. What does what, what's free, what's enterprise, what's promised versus what's already working. The figure that frames the section: 89% of hotel chains plan to expand AI use in the next twelve months (h2c AI Study, 2025). The question the section addresses is with which tools, and by what criteria.

Section G · EU Regulatory Primer · 25 May 2026

What no competing playbook tackles: what the AI Act concretely means for a hotel, a DMO, a tour operator. GDPR reduced to its operational essentials. NIS2 in the parts that actually affect you. No decorative legalese. European regulation can be read as a brake or as a global reference point. American buyers working with European clients are living proof that the second view is the right one.

Section H · Three Steps to Start · 1 June 2026

A ninety-day operational sequence. Audit, first deployment, orchestrated state. No theory — a sequence made of things to do. It's the piece I get asked for most often. It's free, and it includes the points where the typical sequence tends to break down.

Section I · Case Studies · 8 June 2026

Six documented cases, with real metrics. A destination, a boutique hotel, a tour operator, a museum, a winery, an outdoor tourism operator. What they did, in what order, what they paid, what they got, what they'd do differently. Real operators, real numbers, lessons that transfer.

Section J · A Glimpse into the Future · 15 June 2026

The long view. What happens to the sector between now and 2030. Nobody knows precisely. But some signals are already strong enough to be worth reading now, even if the conclusion stays provisional. Native AI trip planning, agentic booking, advertising versus transactional business models, redistribution of value among OTAs, direct suppliers and new AI-first platforms. A section that asks questions more than it gives answers, and that honestly separates the probable from the desirable.

Section K · FAQ and Bottom Line · 15 June 2026

The close. The questions that remain, in FAQ form. And the bottom line on one page, without excuses: what to do on Monday morning. It comes out alongside J because it closes the book.

What you'll find and what you won't

For every claim I make, there's a source. Number, year, organisation that published it. If a piece of data has no source, it isn't there. For every section, a camera flip: the moment in the text where the same reality is rephotographed from an angle that changes what it means. Sections close with a provocation, not a summary. Summaries AI can handle. Provocations require having actually worked on the subject…

What you won't find: wide-angle predictions, digital-humanism rhetoric, exhortations to "embrace change," generic upbeat closes. Zero strategy theater. If you're looking for confirmation that using ChatGPT is enough to count as innovative, this isn't the right book.

A personal note

I've been writing about AI in tourism for a few years now. I've seen everything. Projects that worked because there was substance underneath. Projects that failed because the foundations weren't there, and AI was called in to fill them — which it doesn't know how to do.

What I've learned can be summed up in a sentence I've been repeating for a while: without data, there is no AI. It's the point of Data-Driven Destinations, it's the silent premise of every section of this playbook, and it's why the first thing I ask anyone who contacts me isn't which tool they're evaluating, but what they actually know about their own customers, their own channels, their own offering.

The AI Tourism Playbook isn't a book for consultants. It's for people who actually manage a tourism organisation, have a real budget, and need to decide on Monday morning where to put the first three hours of AI work this week.

If that's you, read it. If instead you're looking for yet another article reassuring you that you can wait another year, there are plenty of those around, and they're all free.

Open the playbook →

The first three sections are already online. The next one comes out Monday 4 May.