No Show

Will AI Kill the OTAs?

• Jeff Borman and Matt Brown

Share episode

AI's true impact on travel remains deeply speculative, with companies scrambling to stake claims while few concrete applications deliver true transformation. But everyday uses like dynamic pricing, translation tools, chatbots, and trip planning assistants like ChatGPT show AI is already here.

We talk about the staggering potential of agentic AI to automatically rebook flights, arrange rides, or sign up for the best rewards programs. AI will directly challenge Online Travel Agents, whose dominance rests on aggregating inventory and managing transactions at scale. Yet AI’s iterative, user-driven planning capabilities expose the weaknesses of OTAs in handling complex, personalized itineraries.

How will the online travel booking sites survive? What will happen to marketing and SEO over the next 2 years? What's a neural travel marketplace? Should we ignore Public Enemy's sage advice and believe the AI hype?

Matt Brown:

Hi everybody. It's no Show. With Matt Brown and Jeff Borman, you will find no bigger boogeyman in 2025 than AI. And, of course, as with any formidable transformative technology, there are a lot of people claiming expertise, prognosticators, sage visionaries and world-weary clairvoyants, all angling to sell themselves as consultants on the future, and travel definitely has its fair share of these profits. Ai is all superlatives the biggest, the most revolutionary in the history of mankind but is anything actually happening with it? And we're here to tell you. You know, we don't know. Nobody knows anything other than AI, like all tech products, is about efficiency and data collection. These are all just guesses about how AI is going to go down for hotels, for travel agents, for airlines, for everybody. Still, we wanted to come here today and talk it out in a kind of 101 session, sort of a state of the union, about what is happening in hospitality with AI right now, and I think the funny thing is we were talking about this.

Matt Brown:

Ai, as with a lot of other parts of our lives, is already here. You know voice assistants and chatbots and facial recognition. They've been around forever. Hello clear Airlines are using dynamic pricing algorithms. Translation tools are now de rigueur in airports and hotel lobbies and hospitals everywhere. Even over the last six months, I've run into I'm sure, jeff, you have too. I've often run into people who are just casually using it and seriously using it as a trip planner, and they're just doing that with ChatGPT and Copilot and whatnot. So companies everywhere are scrambling to own that.

Matt Brown:

So where does this leave travel? Right, because many of the warnings about the future of generative AI are focused on platforms like ChatGPT and the next iteration of Google, and that's because I think people are seeing those now on a daily, sometimes hourly, basis. Companies should be worried, right, maybe they should be more worried than they are. I don't know. We need to kind of look under the hood of what a lot of travel companies are doing, but I think they need to be worried about advancements in mobile phones operating systems. People aren't going to go to a travel site and use their AI bot. They're going to go to the AI bot that's on their phone already and goes to all the travel sites. Again, it's about efficiency. So if Chrome is on your phone, you're going to click on Chrome and it's going to take you away to a magical world. You're not going to want to go through any other hoops. That's how tech has run for 10,000 years and that's how app tech has run for the last 15, right.

Jeff Borman:

The $20 trillion that AI is supposed to unleash and contribute to the global economy over the next five years. Most of that's still going to come through operational efficiencies. That's not what this conversation is necessarily about. The subject of AI can touch anything imaginable, and I think, matt, where we're going to try to focus is on the commercial aspects of travel, not the hotel operational aspect. This isn't about what can AI do for check-in or guest room tech, monitoring the HVAC for optimized use or sustainable lighting controls, touch-free experiences that are more hygienic, and multi-sensory, immersive three-dimensional navigation throughout a property. These things are unlimited. The cool part about the AI subject is that it is still only limited to your imagination, but I think what we're going to try to get to here is what's there today, and particularly for commercial leaders and if you are curious about that, I would encourage you to check out an article on Autonomous.

Matt Brown:

And is the world's first true AI powered hotel in Vegas, of course, you know you can see the marketing language come a thousand miles away. It's revolutionizing hospitality through proprietary AI technology. I love tech, loves revolutions.

Jeff Borman:

So they're always also written by a bot.

Matt Brown:

Totally so. Yeah, the robots have already started the revolution. It's, I think it's behind Mandalay Bay, so it's fairly near the airport, kind of Decatur-ish if any of you out there kind of know Vegas. I can't tell how much of it is hype versus how much of it is going to be real. But they say look, you can customize your stay by selecting preferred amenities and you can determine your desired level of interaction. I think one of the interesting things about the press release was that they say, with guest consent, the AI will gather insights on you before your first stay by analyzing publicly available data. By analyzing publicly available data, and to further enhance personalization, guest will complete a short, gamified I'm sure onboarding questionnaire before arrival and it will ensure a tailored, seamless experience. They love those Tech loves, tailored and seamless things as well. From the moment that you check in, I'll be very curious to see A what the rates are. We got on the site a little bit but I didn't go and look and see what it would take to book a room.

Jeff Borman:

In the world of agentic AI. It would have booked it for you and you didn't even realize.

Matt Brown:

You know most people you know just to kind of get a baseline understanding here, and we needed this as much as anyone. There's AI and then there's agentic AI. The main difference between kind of general artificial intelligence and agentic artificial intelligence lies in their autonomy and their ability to act independently. In general, ai can perform tasks and make decisions that are based on the data and instructions it receives. So the chat GPT model is this hey, chat GPT, give me a summary of blank, or rewrite this, or go give me a secret about this town or this. Who knows Agentic AI? On the other hand that's where we start getting into 2001 and Iron man territory it possesses a greater degree, a much greater degree of independence and can proactively make decisions, take actions and learn from experience without constant human input, and that is both very interesting and a little Terminator scary.

Jeff Borman:

An example here, Matt, of gen AI or generative AI would be kind of an elevated chatbot that answers frequently asked questions about a flight schedule, or a system that recommends hotels based on your past searches.

Jeff Borman:

Specifically, generative AI could be creating content like a personalized itinerary or marketing materials if you're selling. An example of agentic AI would be a step beyond that a system that monitors flight status, knows what flight you're on, automatically rebooks you on the next available flight if yours gets delayed, even arranges for a hotel stay if you need it, then sees that you've landed, orders the Uber on your behalf to arrive at the exact time it would take you to get from where your phone's located over to the nearest rideshare door, get the nearest rideshare pickup spot and it's all happened for you without you doing anything. Agentic is quite a while off, but pretty cool to think about. Some think agentic AI that, by enabling a Google agent or the like to make a purchase on your behalf, would be the end of the checkout page forever, and that we're entering a new phase of digital commerce where intelligent software agents do more than search or respond they actually act on our behalf.

Matt Brown:

AI machines may better manage my payments. They will select payment with the credit card that most rewards me for that type of transaction. Maybe it signs me up for one that I don't have to get a deal that I would have passed on because of the signup process. That's, I think, a huge one, right? It's like if you were to the number of times that even just that little bit of friction as far as like, oh, I don't want to give them my data, I don't want to get hooked into another subscription, another card, but if, if, if your robot helper said I'll take care of all that right now, and if you tell me you want to cancel in three months or whatever the trial period is, I'll do that too. Now everybody wins and you can go off and do whatever you want to do and not have to sit and fill in all your stuff in the blanks.

Jeff Borman:

The scary part to that is that once my authorized agentic AI is interacting with a retailer's agentic AI selling and you've no humans involved in the buying or selling process, just a pair of machines, the potential for fraud is unlimited, totally, totally, because it's all going to happen under your nose and it's not going to happen with monitoring.

Matt Brown:

They'll say there's monitoring. The companies will say, well, we have, you know, we'll have. They'll probably even make a feint towards human supervision, but I think we'll, as we've seen with a lot of other technology, that human supervision can't monitor or control everything that's going on in the system Developers and project managers and QA people who are going to be responsible for billions, trillions of transactions, and that is just going to be crazy. So all this leads to a question of where do the OTAs that we've known and loved for 25, 30 years fit into this new world order? And what OTAs like Bookingcom, like Expedia, actually excel at is single-term, single-item booking One hotel stay, one flight, one experience the classic sort of search, detail, price, compare, book workflow that has dominated travel for decades. It's weird to think that it's dominated for decades because I think if you're of a certain age you can kind of remember a time before that, but it has been the way that we communicate for a quarter of a century at least.

Jeff Borman:

Personally, I don't see the OTAs as being threatened any more than any other party here. I can understand that philosophically, but the scale is unmatched. Booking commands almost 70% of the European hotel OTA market.

Matt Brown:

It's bookingcom right, bookingcom right.

Jeff Borman:

And Expedia's got another 11, so you've got over 80% right, I?

Matt Brown:

had no idea. I had no idea that bookingcom is almost 70% of European hotel.

Jeff Borman:

In the US, the two together make up 93% of the OTA leisure market. Right, so the skill isn't the result of some breakthrough tech, it's operational mastery at a brutal, brutal scale. And they solved the hardest problem in travel from 25 years ago 30 years ago which was aggregating a fragmented inventory while handling the payments, the cancellations and the customer service across thousands of suppliers. Right, they figured that out. That's why they're in a position during the day, 80% of travelers visit OTAs to research and compare prices. That probably is under a bit of threat, but you still have to go book something somewhere. Now, where they are potentially at threat is if you try to plan a very complex week-long itinerary using an OTA that already fails. So it's not like I think they're going to lose that market either. When we had Heather Heverling from Audley, one of the things that we talked about about the future of AI on that personalized travel that they do and that luxury travel advisors bring, an OTA already does not do that well, that's why their market continues to thrive, but AI might be able to start iteratively doing better right, even just in chat GPT that's probably not the best source for planning an itinerary, even if you're using that for that purpose. You just kind of keep refining and refining and refining your search and the results and you refine a little more and you get a little better and better and better and better. That iterative process is not something that an OTA does very well. Process is not something that an OTA does very well, but it is something that AI tech brings right into the mind of the user. However, even today, we're getting back to that kind of dreamy. What could AI become? Even today, large language models break down under that complexity. Openai's most advanced model has a 10% success rate on complex travel benchmarks. That's not enough for a human to go back to it. Earlier LLMs managed less than 1% accuracy compared to a human at 100%. You're getting exactly what you wanted. It just takes your time.

Jeff Borman:

Travelers don't start their journey in OTAs anymore quite to the frequency they did before. Almost 90% of travelers turn to social media for inspiration and 75% on social platforms, and I think you'll find Expedia had a very cool and it's probably more PR than reality, but a very cool thing. They offered up about a month ago where you can book through Expedia an itinerary that it creates with its AI interface into reels or in TikTok. So somebody sends you, matt. Here is a phenomenal vacation reel. This is what we just put together. Here's your flight and our experiences. Here's the hotel and the beach side stuff. I got this cabana for dinner, that restaurant. Expedia is now putting that together into. I saw a reel. It is now a bookable itinerary.

Matt Brown:

Okay, that's pretty good.

Jeff Borman:

That's pretty cool and that exists now. Now it doesn't have the scale there aren't 10,000 of these yet, there aren't 10 million of them yet but the concept is live. It's not just a concept, that is live.

Matt Brown:

Because not only does it sell you on the trip, but it's a great thing for you to share with friends, like, should I do this or should I do this? And then, if you actually choose to do it and people ask, oh hey, where are you going? Oh well, this is what I'm going to do. And now it's this packaged little pre-flight reel that you can share with family and friends and say look at all the cool things I'm going to be doing.

Jeff Borman:

Traditional search and advertising strategies are giving way to a new era of AI. First discovery, Matt, you mentioned this at the beginning. Where anybody who's been on Google in the last six months, you will notice that the AI answer comes before the SEO or the search answer, the keyword, the old school keyword answer. A study by Epsilon showed that 94% of marketers reference using AI today, 100% plan to in the immediate future. Yet only 41% of those respondents say AI is having an impact on marketing performance today.

Matt Brown:

You know, with all this talk about efficiencies, the operational efficiencies should also be available to marketers. Short-term, we have to care about the marketers. They need love too. They're going to use AI to power chatbots, voice recognition, creative ad generation, and these are all very reasonable applications in the present, but I think what everybody really wants is a way to leverage AI to improve customer shopping experience.

Jeff Borman:

I think true adoption just isn't there yet at scale. But that same Epsilon study showed that more than half of marketers feel like they are already air quotes here mature in AI adoption. But I can't believe that for a second. Considering when I ask hotel marketers whether it's a hotel level or a chief of the biggest companies, and I ask this question all the time how are you using AI? I get almost no real answers. And hotel marketers that are engaged in AI related activities. The largest improvement, they say, has been improvement to the return on ad spend, roas. When I read that, I hear affirmation bias, because I have not seen any scalable improvement in ROAS over the last two years. In fact, it's actually been in decline because cost per click has been rising dramatically. Marketing costs are simply elevated. So the efficiencies that marketers are supposed to be getting from AI at best right now it's being chewed up simply because it costs more to do the same thing it did yesterday.

Matt Brown:

That's got to be so frustrating for marketers. Finally, after two decades, after being grilled constantly about what are we getting for our ads and then they have to kind of look around at each other and make up answers because nobody really knows they finally have been able to get a system down where it's like they can answer those questions with with a decent amount of faith and keep their job for another year. And now all of that is going out the window. So I I will light a candle tonight for the travel marketers of the world.

Jeff Borman:

You know what, though? I will not. No, because that candle was supposed to be, because AI was going to take their job, and really what we're talking about here is it's nowhere close.

Matt Brown:

One of the ways that helped us understand a little bit of this was a great paper from Focuswire that came out recently called Four Scenarios on the Future of Agentic, ai and Travel Good title, and it essentially goes through these four different models. And the first one, the kind of basic one, is essentially augmented booking. So it's kind of what we have now, but it's a little more conversational, it's more intuitive and knows your past past days and then makes recommendations Great. The second scenario, now raising the stakes a little bit, is an AI travel concierge. So you put it in the system, you talk into the system and say plan a trip to Portugal in September, it's going to be a week, give me a mix of social and coastal experiences, I have $6,000 to spend. And then your AI assistant picks the best bidder on its OTA shortlist and the one that offers the most suitable travel itinerary flights, all of it. You say book it and then your calendar is automatically updated with all the details.

Matt Brown:

A third scenario is one in which the brands get involved and you open up, say, a Hilton app or Marriott app and you chat with their AI bots about a trip to Philadelphia or Chicago for a conference and the AI suggests very specific options like this hotel connects directly to the convention center and this one has an upgraded executive lounge, which you've rated highly in your past days and you get this personalized package that includes a preferred room type, an airport transfer, breakfast experience, all the stuff and the entire transaction occurs within the Hilton ecosystem and you're guided along very quickly to a satisfactory resolution. The fourth scenario, though, I found particularly interesting. It's the big kahuna, and they call it the neural travel marketplace. You tell your personal AI assistant, your Jarvis, I need to be in meetings the second week of June in this town, build me a suitable itinerary, and your AI robot interfaces directly with the AI agents from airlines and hotels. Robots talk to robots with the AI agents from airlines and hotels, robots talking to robots, and each, autonomously, is authorized to complete transactions. No humans, just instantaneous negotiations that follow your preferred seat positions, your historical sensitivity to hotel room noise, what way you want to face, what time you want to get up, all of it and the robot takes care of the whole thing and you press a button maybe, or at least give some kind of authorization at the beginning or end, and you show up at the airport a month later and take off.

Matt Brown:

And in this landscape ruled by AI agents, otas could fade away Consumer marketing would fade away. Marketing budgets would go to AI agent referral fees, because none of this is going to come for free or cheap. It's going to come to these AI agent referral fees that ensure preferred access to your AI assistant, and those will be owned by big tech giants. They'll probably be developed by startups, but, don't worry, all those startups are hoping that in five years they get bought by Google because their business model is so good or because Google wants to neutralize the competition.

Jeff Borman:

Matt, would you like to ask me a mystery question and I will let you know what chat GPT thinks? The answer should be yes. Make it personal, though. Let's see if it can get me right.

Matt Brown:

Okay, Next year you want to go on a dream trip to a place you've never been before. You have two weeks and let's give you a healthy budget for this one. You deserved it. Let's give you $20,000. I know that's not healthy to some people. It's healthy to me. Where should you go?

Jeff Borman:

Okay, well, matt, I know my answer.

Jeff Borman:

Well, chatgpt is responding for me on my behalf. Thankfully it does not yet have. Well, we'll find out. Maybe I'll be a fan of this agentic thing. We'll see what it says. My answer would be either New Zealand tough on 20K though I like to ride in the front of the bus, you're going to spend 20K just for Heather and I to get there and back in the front. But New Zealand, south Africa never been. Croatia could probably do that, definitely could do that on 20K. Kenya, tanzania those are my big four. All right, let's see what it says. Japan ultra modern meets agent. Let's see Helicopter ride over Tokyo budget 14 to 18,000,. Let's see Helicopter ride over Tokyo Budget 14 to 18,000. Private tea ceremony in Kyoto.

Jeff Borman:

Ok, option number two here we go. South Africa plus safari, cape Town and wine country. Matt, this is exactly what I want. Number three Bali and the Komodo Islands. I've been to Bali a dozen times, so no, thank you, but it's not a bad suggestion. Patagonia, buenos Aires, argentina. I absolutely love Argentina. We just did a Patagonia trip about five years ago. I probably wouldn't do it again, but I would recommend it. So, solid answer, chat GBT, amalfi Coast plus Sicily. I literally got back from Sicily two weeks ago and I've been dreaming about going to the Amalfi Coast. Fucking hell, man. That's impressive.

Matt Brown:

It's impressive.

Jeff Borman:

This knows nothing about me.

Matt Brown:

And in six months. You know, what we should do is we should come back here in six months and ask it the same question, because it'll learn more about you.

Jeff Borman:

It will. Right now. It only knows how to make stupid memes. That's all I ask of it.