PROVIDE INVISIBLE HELP
Think Siri or Alexa: There’s a growing demand for digital assistants that can understand spoken requests and perform tasks, such as scheduling a meeting, calling a friend, or making a payment.
WHO IS MAKING MONEY
Startups that develop machine learning for existing digital assistants stand to make the most money, while some newcomers are creating their own app-based assistants.
HOW TO DO IT
This new breed of artificial intelligence is built on speech-recognition programs and other quickly evolving technologies. If you have a smart group of programmers eager to take on a sophisticated topic, you could develop and sell a standalone app, or license it to third-party companies for a fee. (Take NextIT, based in Spokane, Washington, which licenses its Alme platform to businesses and also builds custom digital assistants for individual companies.) Or you could sell your whole company: Amazon, Apple, Google, and Microsoft have all recently acquired machine-learning startups—some for hundreds of millions of dollars—to power their own digital assistants.
WHO’S SUCCEEDING
A number of startups are making standalone digital assistants, including the Santa Clara, California–based SoundHound and the Mountain View, California–based EasilyDo. And some entrepreneurs are serial innovators: In 2005, William Tunstall-Pedoe launched what became Evi Technologies, a Cambridge, England–based software maker that eventually created its own digital assistant. In 2012, he sold the company to Amazon for a reported $26 million; now, the e-commerce giant uses Evi’s technology to enhance its own digital assistant, Alexa. Tunstall-Pedoe, who left Amazon a year ago, is considering starting another A.I. business. And he sees plenty of room for new A.I.-related startups to find “massive” success: “It’s a super exciting space.”
THE RISKS
Big, big-name competition. Since the tech giants are already making digital assistants, yours needs to significantly improve on what’s widely available. “It would be difficult to create a standard, Alexa-like product unless you have very unique technology,” Tunstall-Pedoe says.