Using AI to plan a holiday isn’t particularly unusual anymore. Ask ChatGPT where you should stay in Tokyo, what you can do in London for three days or to put together an itinerary for a trip through Italy and it’ll happily oblige. The problem comes afterwards, when you still have to open a collection of websites and apps to check the suggestions, compare prices and make the bookings yourself.
Trip.com and Mastercard are working on cutting out some of that work. The companies have announced a new AI-powered booking system that will allow Trip.com’s TripGenie assistant to search for travel products and, with the customer’s permission, complete the purchase too. Network International is involved on the payments side, while Mastercard is providing the technology that allows the AI agent to transact.
TripGenie has been around for a while and can already help travellers search for destinations, flights, hotels and things to do. What Mastercard is adding is the ability for the assistant to go beyond finding something and actually buy it.
In this case, that’s essentially what “agentic commerce” means. You might tell TripGenie that you’re looking for an attraction during a weekend in Dubai, give it your requirements and budget, and then allow it to complete the booking once you’re happy with what it has found.
Mastercard says its Agent Suite for Merchants can handle discovery, research and consumer-authorised purchases inside a company’s existing online store. Trip.com will initially use it for selected travel services such as attractions and other add-ons, with a phased commercial launch expected in early 2027. There’s no confirmation yet that South Africa will be one of the first markets to get it.
AI has already made itself useful in travel
Planning a complicated trip can involve a ridiculous amount of admin. Finding flights is one thing, but then there are hotels, airport transfers, restaurants, attractions, public transport, weather, visas and whatever else you’ve decided to squeeze into the trip. There’s also no shortage of choice, which sounds great until you’ve spent two hours comparing 34 hotels that all appear to be five minutes from the same train station.
Research into AI and tourism has been looking at this problem for years. A 2026 chapter published by Emerald identifies personalised itineraries, travel recommendations, booking assistance, real-time flight information, customer support and concierge services among the areas where AI is already being used. Virtual assistants can also handle fairly mundane jobs such as changing bookings, answering common questions and helping with cancellations or refunds.
Research published by EY looks at how generative AI can be used to personalise travel recommendations while reducing some of the work involved in researching and booking a trip. The Youth Tourism Summit was discussing many of the same applications several years ago, including trip planning, customer service and personalisation.
The technology has moved on considerably since then. Instead of only asking an AI where to go and what to do when you get there, companies such as Trip.com are now trying to give it access to the part of the process where money changes hands.
Paying for the trip raises different problems
Travel information doesn’t stay still for very long, which becomes more important once an AI is able to act on it. A hotel can sell out, an airline fare can disappear, an attraction can change its opening hours and a room that cost R2,000 when you started looking can cost considerably more by the time you decide to book it.
There’s also a difference between an AI recommending the wrong restaurant and booking the wrong flight. In one case you might have a disappointing dinner. In the other you could be trying to convince an airline that a piece of software accidentally sent you to Frankfurt on the wrong Thursday.
The 2026 Emerald research raises privacy and security as concerns as AI becomes more involved in tourism, particularly because personalised systems depend on information about the customer. Mastercard says purchases made through its system remain consumer-authorised, so the idea isn’t that TripGenie gets access to your card and starts making travel arrangements without asking.
For the assistant to be genuinely useful, though, it’ll probably need to know quite a lot about you. Your budget is easy enough, but it helps if it also knows which airlines you prefer, whether you’ll tolerate a connecting flight, the kind of hotel you usually choose, what food you eat and whether you’d rather spend Saturday in a museum or on a beach.
None of that information is particularly alarming on its own. Put enough of it together and you’ve built a fairly detailed account of someone’s habits, preferences and movements.
Who decides what TripGenie shows you?
Online travel booking has never been completely neutral. Results are ranked, airlines sell different fare classes, hotels pay commissions and booking platforms promote particular offers. Most of us understand that to some degree because we can see the results page in front of us and keep scrolling if we don’t like what appears at the top.
An AI assistant changes the presentation. Instead of looking through 100 hotels, you might ask for something quiet, central and under R3,000 a night and get three recommendations back. At that point, how those three were selected becomes rather important.
If TripGenie tells you that it has found the best hotel for your trip, you’d probably like “best” to mean the option that most closely matches what you asked for. The answer becomes less useful if commercial arrangements have a bigger influence than your preferences.
Sponsored search results and promoted listings have existed for years, so none of this begins with AI. What changes is how much of the selection process you can actually see when a conversational assistant reduces hundreds of options to a few answers.
South Africa isn’t starting from zero
South African generative AI adoption reached 23.1% of the working-age population in the first quarter of 2026, according to Microsoft data. That doesn’t mean people are suddenly ready to let an AI organise and pay for an entire holiday, but using these tools is becoming considerably more ordinary.
Travel could also be one of the easier places to understand the attraction. Booking an international trip still involves moving between airline websites, accommodation platforms, maps, emails and payment screens. A service that can remember what you’re looking for while doing some of that work for you has a practical use beyond simply being another place to chat to an AI.
There are still plenty of reasons to check its work. Flight dates would be fairly high on my list. Hotel location would probably come next, followed closely by whether the wonderfully cheap room TripGenie has found actually includes a bathroom.
Trip.com expects the commercial rollout to begin in early 2027. If it works as promised, booking a trip could involve considerably less tab-hopping than it does now. I’d still be checking the confirmation email.