Sometimes you just need passion, not loads of market research, to embark on your company’s next project.
That was the case for Kayak, the Stamford, Conn.-based travel search engine owned by The Priceline Group, when the company’s head of data science in Boston expressed his love for the Amazon Echo in a LinkedIn post back in December. Simply enough, Kosmas Karadimitriou was enamored with how Alexa, Amazon’s voice service that powers the Echo, proved to be a better virtual assistant than Apple’s Siri or Google Now.
“Although usually I prefer to wait until technology matures before getting on the wagon, with Alexa it was love at first sight,” Karadimitriou wrote.
From there, it didn’t take too long for Kayak to start working on its own app for Alexa, which officially launched on Tuesday, allowing users to ask the voice service about flight times so you can know which time you’re supposed to head out the door, what kind of trips can be made on a certain budget and other inquiries that can be made through Kayak’s website and apps.
For instance, you can say, “Alexa, ask Kayak: where can I go for $400?”
And Alexa will respond, “Kayak’s data suggests that 400 could get you to Los Angeles, San Francisco or Seattle.” And any inquies will be reflected right on Amazon’s mobile app for Alexa. While the Alexa app can give you quotes for hotels and flights, it doesn’t yet include the ability to book them, the company said, but it’s looking into that for future iterations.
In an interview with BostInno, Kayak CTO Giorgos Zacharia says work on the Alexa app—called a “skill” by Amazon—was not the result of market research or noticing that the Amazon Echo was selling like hot cakes. Instead, “it was basically responding to the intensity of one Kayak employee”—Karadimitriou, in this case.
After having initial discussions in December, Zacharia says Kayak’s team started working on the project in early March. That may not sound like a lot of time to work on a product that came out only today, but Zacharia says the foundations for a voice-powered app were already there with natural language processing and machine learning technologies that have been developed for Kayak’s text-messaging service that launched in December and a Kayak bot for Slack that launched in March.
“We realized that many of our users, especially younger ones, spend so much time in text platforms, and we saw how WeChat became like an internet operating system for China so we thought we should be there as well,” Zacharia says.
Matthias Keller, Kayak’s chief scientist, says Amazon’s Alexa team in Seattle also provided plenty of support to Kayak in the development of its Alexa app, starting with onboarding and going all the way through internal testing and certification.
“It was a great opportunity because if you have something that is there for many years like iOS development, you can read all of these best practices on the internet,” Keller says, “but this is so new, and working with Amazon opened so many important thoughts for us on things to take care for,” like when Alexa doesn’t properly interpret a query.