A Watson-Powered Travel Agent From The Founder Of Travelocity

The former chairman of Kayak is betting you’ll want AI to plan your next vacation.

A Watson-Powered Travel Agent From The Founder Of Travelocity
[Photo: Flickr user Nelson L.]

Despite all the apps crowding the category, planning a trip is still a pain. Sure, the booking process is more efficient, and you can find the cheapest flight and the highest rated restaurants more easily than before the Internet upended the travel industry. But the process still requires multiple interfaces, different apps, and in the end you’re still entering the final details into a spreadsheet or note. It’s a tedious project all its own.


But it’s nothing a little cognitive computing can’t fix.

WayBlazer is a new startup from Kayak and Travelocity veteran Terry Jones that’s aiming to solve the travel problem. One of the first companies to team up with IBM for the launch of its new Watson artificial intelligence APIs, WayBlazer is in the early stages of making intelligent machines that act as our travel agents and tour guides.

“People forget that travel is the largest e-commerce marketplace,” say Jones. Online travel booking in the U.S. reached $103 billion in 2012, according to Comscore. “I thought maybe there’s a space a little farther up the value chain, up into the inspiration and dreaming and early planning phase.”

Watson’s unique brand of cognitive computing is especially well-suited to helping people think ahead, Jones says, because “its real strength is structuring unstructured data.” This, combined with things like natural language processing and user behavior modeling, could dramatically streamline the process of planning a trip–or even coming up with the idea to go in the first place.

Unlike Jones’s previous endeavors, WayBlazer is not a consumer-facing portal for planning trips, but rather will start out as a B2B company. Its first client is the Austin Convention and Visitors Bureau, for which WayBlazer will power a web-based cognitive search tool designed to help convince people to travel to Austin.

When it launches publicly, the ACVB search tool will let users type in natural language queries–“I’m in the mood for some live jazz in Austin tonight”–and get a set of recommendations powered by various editorial and social data sources aggregated by WayBlazer with Watson’s help.


This is WayBlazer in its most rudimentary form. That’s no accident. “Watson is not so much programmed as trained,” says Jones. “You have to get the data in and train it how to read that data and deal with that data. We wanted to start with a bounded problem. Austin is a bounded problem.”

Indeed, Jones’s vision for machine-assisted travel planning is much, much bigger. The idea is to build out intelligence about Austin and then stake out new locales to map. Many data points will carry over into new cities, while other details will need to be taught. Over time, some of the learning will become automatic.

“Data is the next big natural resource, but it isn’t worth much until it’s refined, kind of like oil,” says Jones.

Just think of all the data that exists about a city: Everything from restaurant reviews, weather, traffic, and venue check-ins to more nuanced things like how people feel about certain places and events. What’s the overall sentiment of tweets and check-ins coming from the zoo? Where’s the best spot for a trio of 30-year-old women to go on a Friday night?

“What you want to do with your family is not what you do when you’re by yourself, is not what you do with your kids,” says Jones, who began his career as a travel agent back in 1971.

Then there’s the other side of the equation: You. Early implementations like the ACVB site aren’t personalizing results, but it is absolutely Jones’s intention to do so. Once a user logs into a system powered by WayBlazer, they will be giving the system–consensually, of course–their identity, the most important data point of all.


This, in theory, will let WayBlazer tap into a trove of actionable information, from Facebook likes and the sentiment of your tweets to travel and commerce data that’s already being collected by hotels and airlines.

“My favorite airline knows that I always sit in seat 4E, but they don’t do anything with that data,” says Jones. “The hotel knows what kind of room I want, but they don’t do anything with it unless I explicitly tell them.”

Whether we like it or not, there’s a ton of information about our behavior sitting in various databases in various formats, never being linked together. With our permission, Jones would like to start piecing it together and making use of it in a way that only Watson’s machine intelligence can.

“It’s all dark data,” says Jones. “We’d like to shine a light on that dark data and actually help people use it and take this digital exhaust and refine it.”


About the author

John Paul Titlow is a writer at Fast Company focused on music and technology, among other things.