The agents are coming

Smart agents are on the horizon and they’re arriving faster than you think.

Amazon Echo’s Alexa

Very soon, anyone buying a high-end smartphone will expect the handset to obey verbal instructions, respond to queries with spoken answers, and even anticipate the needs of the user.

That’s partly because devices like the iPhone and the Samsung Galaxy S8 already have ‘smart agents’ built in, going by names like Siri and Bixby.

But you don’t need a cutting edge phone to experience the cutting edge of artificial intelligence. Numerous apps are starting to build in smart agent functionality.

Arthur Goldstuck Arthur Goldstuck
An early example is SoundHound, the app that allows users to identify any piece of music by playing it or even humming it with the app open. It’s the poor relation of Shazam, the most popular app for this purpose.

SoundHound needs to find a way to attract attention away from Shazam, and it may have found it via a new smart agent function. The user opens the app, and starts a sentence with “OK, Hound…” and then gives an instruction like, “Play the Beatles”. The app searches its archives for Beatles songs, produces a playlist, and begins playing it.

Intense competition

Most people, however, will have their first smart agent experience via a branded device in the form of a smart speaker, like Amazon Echo’s Alexa, Google Home or Sony Agent.

Last year saw this new category become a market segment in its own right, with six million smart speakers sold, according to global analyst firm IHS Markit. By 2020, annual sales are expected to reach 59 million.

“We will see prices drop extremely fast on early products,” said the firm’s principal analyst, Paul Gray, speaking in Lisbon at the recent IFA Global press conference, an event that previews Europe’s largest consumer electronics show, IFA, taking place in Berlin in September.

“Alexa has snatched the lead due to perfect timing of the market, but competition will intensify from other platforms and devices.”

With Microsoft’s Cortana also in the mix, most of the big guns of software and cloud services are showing their hands. But this isn’t only about controlling platforms for the sake of those platforms, said Gray.

“It’s about the power of the network, and the device allows you to access the network. The power of the network, in turn, reinforces the device. The more people talk to Google Assistant, the more it improves recognition of accents and minority languages, for example.

Players that are early in this will have better trained neural networks, which means they will outperform the others,” he noted.

Computer makers naturally want a piece of the action, but are also looking at the longer term potential.

“What I see in AI today is like the internet back in 1995, when we were just using Netscape to go online and research some information,” says Maverick Shih, president of the Bring Your Own Cloud smart products division at global computer maker Acer. “Then came a bookstore called Amazon, services like weather, and suddenly we had the all–encompassing internet services you have today. That’s about to happen in AI.”

Speaking during the Next@Acer global press conference in New York recently, Shih argued that that the current high-profile applications of AI, like smart assistants and chatbots, were merely low-hanging fruit. Its potential uses are almost unimaginable, but the clues are already available.

“Right now, on Google Photo, instead of searching by remembering when you took a pic, you can type in a keyword and it identifies the location or the subject in the photo. That’s happening in my own library. Imagine where it could go.”


  • Arthur Goldstuck is the founder of World Wide Worx and editor-in-chief of Follow him on Twitter and Instagram.


    sponsored by
    sponsored by