How Artificial Intelligence Could Reshape How Travelers Book Hotels
Skift Take
It’s easier to talk about people, companies, or events than to talk about ideas. But one idea worth discussing, despite its complexity, is how artificial intelligence could reorder hotel distribution.
Some researchers are wondering if artificial intelligence could handle some of the more complex tasks of shopping and haggling. Could new algorithms and processes shrink the role of travel search engines and comparison apps? Could the cost of bringing buyers and sellers together shrink thanks to technical innovations?
“What is cumbersome and inefficient is the way we, as consumers, interact with travel businesses,” said Humayun Sheikh, co-founder and CEO of Fetch.ai, a startup based in Cambridge, UK. “We need a shift of the technical architecture in the middle.”
Sheikh knows a thing or two about these ideas. He was an investor in DeepMind, an artificial-intelligence company that Alphabet acquired as a sister division to Google. He’s helped raise more than $21 million for Fetch.ai, his new project.
We asked Sheikh how he would explain the new concept to one of the co-founders of Airbnb if he happened to sit next to one of them on a flight by chance.
“That would probably not be a very good conversation because the whole concept here is that you don’t need an Airbnb,” Sheikh said.
Join Us at Skift’s Online Travel and Distribution Summit on February 17
“Today, if you have a room in your house and you want to rent it, you go to Airbnb or a competitor, and it aggregates the supply and demand,” Sheikh said. “But tomorrow, the consumer may have a piece of software made by us or someone else that can look for all the rooms available via an alternative system that replaces a centralized aggregator like Airbnb.”
Reimagining Hotel Distribution
Sheikh said artificial intelligence could smooth out how consumers book hotel rooms. Details vary on how the distribution changes might happen, but some broad concepts are being outlined now.
In one vision, consumers could use an app that they set to automatically book hotel rooms when available for specific dates, based on their preferred rate, quality, and location choices.
This “app” might not be something consumers download specifically for travel. Makers of smartphones might include the AI-powered “app” as a default utility for booking services like travel but also parking spaces, hairdresser appointments, and other products and services.
For their part, hoteliers could use software to sell hotel rooms when they receive acceptable requests from the consumers.
The cost savings for hotels and consumers, when compared with today’s distribution model, could be up to 10 percent, after factoring in costs to run the new system.
“You wouldn’t pay as high fees to aggregators because this is a cheaper way to distribute,” Sheikh said. “The system we have enables a hotel owner to launch a piece of software without a set-up cost, as we have more of a pay-as-you-go approach. The hotelier’s software will interact with the software that consumers will use to shop.”
Amadeus Deal
Today, middlemen, such as wholesalers or online travel agencies, manage these negotiations on behalf of both sides, matching consumers with rooms. Under this new model, software controlled by the consumers and the hotels handle the transactions instead.
Earlier this month, Fetch.ai signed a deal with travel tech provider Amadeus which gave it access to more than 700,000 hotels.
Software programs can access Amadeus’s servers to fetch a hotel’s name and location and other information needed to complete a traveler’s booking without using an online travel agency or metasearch site. Fetch.ai is seeking other inventory partners, too.
Hoteliers won’t try to sell all their inventory on this new channel powered by artificial intelligence. It will be an extra channel, adding to the mix of online travel agencies, bedbanks, wholesalers that work with ethnic travel agencies, and tech companies that bundle together packages for tour operators to sell.
Artificial intelligence could bring the underlying business model back to something like a customer calling a hotel to book a room. Only a customer wouldn’t pick up the phone. The customer would instead have a piece of software do the work for them. The hotelier would have their own software, instead of an employee, to handle the request.
Nerds will want to know that this process is based on the computer science concept called decentralized, multi-agent systems. The process involves letting specialized computer programs do particular tasks on their own, with the help of guidelines that humans give them. The software programs talk with each other. They can aslo collectively grasp tricky calculations without the need for an all-knowing, centralized database or a master puppeteer pulling strings.
Blockchain, or distributed ledger, technology is also part of the process. This use of blockchain is unrelated to the bitcoin currencies that are also powered by blockchain. A few travel companies, such as Travelport and tour operator TUI have experimented with blockchain as a kind of secure database.
Promises of Enhanced Data Privacy
Why would consumers want to learn something new, like using one of these programs? Saving money is one factor we’ve mentioned.
Privacy protection is another of the promised appeals.
“Your data doesn’t leave your piece of software,” Sheikh said. “Fetch is not a data aggregator, and we’re not going to take your data and upload it anywhere.”
The privacy controls make Fetch.ai’s model stand out from today’s world. Today, aggregators that want to provide personalized service, such as recommending a few hotels at the top of your search results that they think you’ll like, need to upload, analyze, and store customer information.
Under today’s model of “personalization,” once the aggregator, or middleman, knows you are willing to spend up to $300 for a hotel room, it will use that information to influence its future recommendations to you for months to come.
But under the model Sheikh is sketching out, the customer’s preferences never leave the software.
This privacy concept isn’t unique to Fetch.ai. The idea of an internet future where a person’s data stays with them instead of gets sucked up by tech titans has been touted by Sir Timothy John Berners-Lee, a computer scientist known who invented the World Wide Web in 1989. Berners-Lee has championed a move to new technologies that would give individuals more control over their data, such as at an Amadeus T3CH conference in March 2019 and earlier this month in the New York Times.
Berners-Lee calls the concepts “pods,” or personal online data stores. Let’s leave the technical details aside. For brevity’s sake, imagine something like a whirring gizmo that stores all your key data, such as your shopping habits, preferences, and key purchases. It’s something like a digital vault.
In the new internet that Berners-Lee envisions, a person could click a button to let their pod to give limited, temporary access to a slice of their data to help complete a transaction, such as apply for a mortgage on a home or book a vacation abroad. A merchant could only use the information for completing the transaction and not for using it to upsell the customer later.
The Fetch.ai concept operates on a similar or parallel notion. A consumer’s preferences remain in their software rather than get sucked up by merchants unless consumers volunteered to allow companies to store their preferences data.
Artificial Intelligence Leaps Ahead
Fetch may sound, well, far-fetched. How much of its pitch is hype?
After all, the travel sector has been hearing about artificial intelligence, or AI, for quite some time. Already many travel companies use artificial intelligence to help automate repetitive manual processes. Examples include customer chatbots for answering routine queries and software that recommends new hotel rates to revenue managers after analyzing changing signals in supply and demand.
Yet every day, the technology gets better. So businesses will one day use artificial intelligence in even more sophisticated ways.
“AI is in the process of abandoning the ‘standard model,’ where a fixed, known objective is given to the machine,” said Stuart Russell, the Smith-Zadeh Professor of Engineering at the University of California at Berkeley, who has worked on new research on how a system can infer human goals and preferences rather than have someone train them on the goals and preferences.
In December, MIT said it has been using a new programming system called Gen to infer people’s intentions based on actions and mistakes. In limited early testing, the system was fairly accurate
In another effort announced this week MIT researchers said they have developed a so-called neural network that learns while performing tasks, not just during its training.
Both developments could someday be relevant to hotel distribution. For example, artificial intelligence could help respond in real-time to changing behaviors by travel buyers and sellers, adjusting rates in sync with shifts in supply and demand.
Problems to Surmount
Fetch’s model assumes that consumers have one set of preferences for, say, hotel rooms. Yet any given person may shop as a budget-minded leisure traveler for one trip, a luxury-minded one for another, and a price-insensitive business traveler when needing to make a crucial last-minute trip to close a deal.
“Broadly speaking, you can have two ways to deal with that,” Sheikh said. “You can have multiple pieces of software searching for different types of travel, or you have just one piece of software that has different skills that you drag and drop into it.”
Sheikh claimed adding skills or choosing a piece of software could be easy and intuitive for a consumer, like how today Amazon Alexa lets users pick skills they want their voice-based internet system to be especially good at recognizing and responding to.
What about contractual and regulatory issues? How would the software that hotels use make sure they only distributed rates that were in compliance with local laws or other contractual agreements, such as rate parity? Sheikh said developers can design safeguards to address such concerns.
A Chicken or the Egg Problem
To build a marketplace, you need sellers and you need buyers, but neither wants to join until you have a critical mass on the other side.
Finding consumers and teaching them to use a new way of booking hotels would need a big marketing effort. As a workaround, Fetch could partner with a player that has already earned the trust of consumers.
One metaphor that comes to mind is Bidroom, a startup in the Netherlands. Bidroom has persuaded about 160,000 hotels worldwide to sell rooms through it for, typically an “activation fee” of a few hundred dollars or euros. Then, after it has proven its effectiveness at putting heads in beds, the startup adds charges. But it won’t charge hotels a commission on the rate, and it claims the total cost of distribution is lower for hotels.
On the consumer side of the marketplace, Bidroom has partnered in Europe with Visa, a credit card issuer that has experimented with offering access to Bidroom’s services to its cardmembers as a free perk. Visa pays Bidroom a fee for the deal. Bidroom has free and paid models for consumers to access its inventory.
So it’s possible Fetch might partner with a company like Visa that has already brought together many “eyeballs” to offer its system as a free perk.
Sheikh wouldn’t be drawn into discussing such possibilities. He said the company is still in the stage of plotting a market strategy.
Yet he expected many hoteliers to find its offer appealing.
“We’re not after the big hotel chains unless they’re interested in us for business-to-business services in making their internal processes more efficient,” Sheikh said. “We’re looking at the small and mid-size operators.”
Sheikh said smaller hotels are increasingly at a scale disadvantage when using the big platforms to sell their rooms on the global online travel platforms. That’s because the giants have bigger marketing budgets and can afford to hire savvy technical talent to game online ad auctions.
So how does Fetch.ai make money while it waits for a critical mass of adopters? Right now, it sells technology to companies in a variety of sectors that want to make operations smarter than what they currently have with their current tech systems.
Like Having a Concierge Book Your Travel For You
Fetch won’t have a monopoly on the idea of creating software for booking hotels and other services. But it aspires to be like the main platform where people who write such software release and use it.
It immodestly thinks of its potential as becoming like Google or the Apple App Store. After all, Google became popular because it was much better at finding webpages than other search engines, and the Apple App Store helps consumers find useful apps. Fetch thinks it will help people and businesses discover the most useful pieces of software infused with artificial intelligence.
“This is the new Internet, which we feel is coming,” Sheikh said. “This is going to be more active and a lot more intelligent than the era of Web pages. But you need a way to find these pieces of software. So we’ll be building a kind of Google for this new internet era. We’ll help you navigate it with the equivalent of search and discovery.”
Well, that’s the idea anyway. There’s a lot of “ifs” in this vision of hotel distribution’s future.
Register Now For Skift’s Online Travel and Distribution Summit on February 17