Learn Tourism - a nonprofit academy

When AI Agents Shop for Travel, Who Trains the Humans?

Written by Stephen Ekstrom | Nov 7, 2026, 10:00:00 AM

Not long ago, I joined my friends Hélène Millet and Ann Cederhall for a LinkedIn Live called Training in a Fast Moving World. We went in planning to talk about airline retailing, AI, and training. We came out somewhere much more interesting: a conversation about what happens to the people in tourism when software starts doing the shopping.

That's not a far-off scenario. AI agents (tools that can search, compare, and book on a traveler's behalf) are already out there trying to make reservations. Some of them are succeeding. Many are hitting dead ends. And the businesses on the other side of those dead ends usually have no idea it's happening.

Here's what I took away from the conversation, and what I think it means for destination teams, frontline staff, and the small businesses that make a place worth visiting.

First, the plumbing (in plain English)

Hélène and Ann live and breathe airline distribution, so I opened by trying to explain one piece of it simply. Call it a small demonstration of how we try to teach first at Learn Tourism.

Airlines have been moving to something called NDC, or New Distribution Capability, a standard developed through IATA. In the old model, travel sellers mostly saw a fare and a schedule. With NDC, an airline can sell more like a modern online store: seat maps, photos, bundles, and offers personalized to the customer or the channel. Bags, seats, Wi-Fi, and lounge access can be packaged into one offer instead of being bolted on afterward. Once you buy, it becomes a single order the airline can manage from end to end.

That's the plumbing. And it matters for one big reason: an AI agent can only evaluate and book a personalized offer if there's a system in place that can produce one.

Hélène added an important caveat, and she's right. A standard is just a messaging format. Plumbing doesn't decide whether water, milk, or wine flows through it. Somebody still has to imagine the product, build the offer, and decide what's worth selling. Ann pointed out that the airlines doing this best are the ones pairing good technology with real investment in the direct relationship with their customers.

Which brings us to the humans.

The AI agent doesn't roll its eyes. It just books someone else.

I live and work on the road, and I'm constantly looking for my next RV park, campground, or host site. When I call a place and nobody answers, I don't wait. I call the next one.

AI agents behave the same way, only faster and without the frustration. If an agent tries to book a hotel, a tour, or an attraction and can't get a clear answer, it doesn't complain. It doesn't leave a voicemail. It moves on to a competitor.

"The AI agent isn't going to complain. They're not going to roll their eyes and put down the phone. They're going to go and book a competitor."

If reservations, sales, and contact center teams don't understand how these agents evaluate offers, hotels, attractions, and yes, airlines will lose bookings. And they'll never see the cost, because a booking that never happened doesn't show up in any report.

Good news for the charming little hotel

Hélène asked the question I hear from many small business owners: if you run a lovely independent hotel on the Mediterranean coast, do you stand a chance in this new world, or does the biggest marketing budget still win?

I think the new rules go back to a very old idea: remove the barriers between your experience and the customer.

In my campground example, the barrier is an unanswered phone call. For a hotel, it might be a booking process with too many steps, rates nobody can make sense of, or a surprise resort fee at check-in. (Don't get me started on resort fees.) It might be a complaint that shows up in review after review and never gets fixed.

AI takes all of that into account when it makes a recommendation. That's a challenge, but it's also an opening. If a traveler is the kind of person who loves a charming, independent seaside hotel, that hotel should be showing up for them. Small properties that are authentic, easy to book, and responsive to feedback can compete in ways that once required a big ad budget.

Ann made a related point that's worth underlining for every destination marketer: when AI is doing the searching, what matters is how often you're mentioned, in what context, and how credible those mentions are. Your website, blog, social posts, reviews, and media mentions are all being read, weighed, and ranked so the best answer for that individual traveler rises to the top. (If that's keeping you up at night, our course The Search Revolution: Building Online Authority in an AI-Driven World was built for exactly this.)

What training should focus on now

Here's the line from the conversation I'd most like DMO leaders to sit with:

"The training that we offer to individuals needs to prioritize judgment, empathy and problem solving over transactional skills."

As AI takes on the simple transactions, the human moments become more valuable, not less. Group travel. Meetings and incentives. Complex itineraries. Visitors who need accessibility features. Recovering service when something goes wrong. That's where people earn their keep.

Hélène summed it up nicely: if AI helps with the transactional work, people are free to focus on the human side. That means caring, knowing the destination deeply, and pointing each visitor toward what's right for them, not just what's easiest to sell.

A few practical places to start:

  • Teach your reservations and sales teams how AI agents shop. They don't need to become technologists. They need to understand that clarity, accurate information, and easy booking now matter to software as much as to people.
  • Teach empathy as a skill, not a slogan. It's not "treat others as you want to be treated." It's "treat others as they want to be treated." That's a skill frontline staff and tourism ambassadors can practice.
  • Keep a human in the loop on itineraries. AI can list every attraction in town. It doesn't always know that two of them are on opposite sides of the city during rush hour. Tour operators and visitor services staff still catch what the map misses, at least for now.
  • Make AI literacy part of workforce development. Destinations have been slow to adopt AI, often because changing how you price or package things isn't exactly a crowd-pleaser with stakeholders. That gap is a training opportunity, not just a technology one. Our AI for Tourism Teams course is one practical starting point.

AI as the bartender

One of my favorite moments in the conversation was about dinner.

Remember sitting at a hotel bar with a glass of wine, asking the bartender where you should eat? By the time you finished your drink, they'd learned enough about you to send you somewhere perfect, maybe around the corner, maybe a few blocks away.

AI is starting to do the bartender's job. Some destinations now measure visitor density at popular spots in real time and adjust what their app, website, and frontline staff recommend, steering visitors toward experiences that are less crowded and often more authentic. Picture asking where to have lunch and seeing where you can actually walk in and sit down, which might send you to the family-run restaurant instead of the chain.

That's a real tool for managing overtourism without immediately reaching for timed tickets and entry fees. But notice what it depends on: someone has to teach the AI what your destination wants to recommend, and in what voice. Do you want to be cheeky? Warm? Buttoned-up? Humans do that teaching. That's a new kind of destination training, and destinations should be building the skills for it now.

The parking lesson

Here's a story I heard recently that I keep coming back to. A destination partnered with an AI company to find out what people were actually asking about it. The number one question wasn't about attractions, restaurants, or events.

It was about parking.

The destination had never seen itself as being in the parking business, so it had never talked about parking. Once it knew, it could start answering those questions, partner with people who solve parking, and build credibility by meeting visitors where they actually were. (Ann laughed, because parking had become one of the hottest topics in her corner of Sweden this past summer.)

The lesson goes well beyond parking. AI can show you the gap between what you think visitors care about and what they actually ask. Closing that gap is a team sport, and it often means collaborating with partners you've never worked with before. It's the same thinking behind building a stewardship team with the staff you already have.

A five-minute exercise to start

If you do just one thing after reading this, try it:

  1. Open Claude or ChatGPT.
  2. Ask it to interview you: "Ask me some questions to get to know me and how I want to work with you."
  3. Answer honestly for five minutes.

That's it. You'll be surprised how much it shapes what you get back afterward. Then try the same idea as a team: what personality should your destination have when AI recommends it? Write it down. That's the beginning of teaching the machine to sound like you.

The two kinds of people (and organizations)

There are really only two types of people: those who embrace change and keep learning, and those who are terrified of it and retreat. Organizations are made of people, so they sort themselves the same way. Some leaders will embrace these tools, make mistakes, and learn from them. Others will try to control as much as they can, for as long as they can, until that's no longer sustainable.

I'm optimistic about which way our industry is heading. COVID showed many people that tourism encompasses finance, sales, marketing, HR, law, and more. Universities are shifting from "how to run a hotel" toward teaching tourism as a whole system. And the students and interns I talk with still want to travel. They just want better careers in the industry than the ones people used to assume were on offer.

At the end of the day, AI and these new retailing systems exist to serve the business. But to do that, they have to serve the customer. A happy customer is a repeat customer, and keeping customers happy still requires people who are well-trained, curious, and empathetic.

Thank you, Hélène and Ann, for another great conversation. You always make me think, and you always make me laugh.

Keep learning

Frequently asked questions

What is an AI agent in travel?

An AI agent is software that can search, compare, and sometimes book travel on a person's behalf. Instead of a traveler visiting several websites, the agent evaluates the options and acts on what it finds.

What is NDC in airline retailing?

NDC (New Distribution Capability) is a standard developed by IATA that enables airlines to sell more like modern online retailers. It supports richer content, bundled extras, and personalized offers, and pairs with an "offer and order" model in which each purchase becomes a single order the airline manages end-to-end.

How do AI agents affect hotels, attractions, and tour operators?

If an AI agent can't get clear information or complete a booking, it typically moves on to a competitor. Businesses that are hard to reach, confusing to book, or slow to fix recurring issues in their reviews risk losing bookings without ever seeing it in their reports.

Can small independent businesses compete when AI does the searching?

Yes, and in some ways it's easier. AI recommendations weigh authenticity, reviews, credibility, and ease of booking, so a small business that removes booking barriers and responds to feedback can show up for the right travelers without a large advertising budget.

What skills should tourism training focus on as AI grows?

Judgment, empathy, and problem-solving. As AI handles simple transactions, the most valuable human work is complex service: group travel, accessibility needs, itinerary judgment, and recovering when something goes wrong. Basic AI literacy belongs in the plan too.

How can a destination use AI to manage overtourism?

Some destinations measure crowding at popular sites in real time and adjust what their apps, websites, and frontline staff recommend, steering visitors toward less crowded and often more authentic experiences. Humans still decide what the destination recommends and how it sounds.

At Learn Tourism, we help destinations prepare their teams, partners, and communities for a fast-moving world, building confidence with new tools and strengthening the human skills no tool can replace. If you'd like to talk about training for your team, let's talk.