If Amazon’s Alexa had just served as an entry point to the company's mega-marketplace, it would have been as boring as the failed Amazon Fire smartphone. But Alexa is way more than that.
Alexa feels like an AI, nothing less. She’s a personal assistant that lives in the cloud and is always listening to you, through a number of available physical devices that use her as their brain. These include the Amazon Echo, the Echo Dot, the Amazon Tap, and Fire TV.
When the first embodiment of Alexa, the Echo, was introduced in late 2014, many thought the device was just a thing you could talk to to buy stuff on Amazon. That’s why Alexa was roundly mocked by tech pundits for the whole first chapter of her life.
But as the laughter died down, people began buying, and loving, the Echo. The Echo is just a thing with some microphones and a speaker; what they loved is Alexa, not her body but her brain. Some people bought two or three of her. People built special niches in their walls for her. The reviews on Amazon were glowing, and still are.
Standing on the kitchen counter, Alexa proved simple and immediately useful. She can read you recipes or the news. She played Spotify or Pandora streams. She made to-do lists for you. And yes, she ordered stuff for you on Amazon.
Alexa isn’t yet as smart and functional as other personal assistants like Apple’s Siri and Google’s Google Now, but she hears you better, she isn’t trapped inside a phone, or inside some platform's garden walls. And Amazon has made it easy for third-party developers and device makers to integrate with Alexa. These early strategic choices by Amazon will amplify the leaps forward in intelligence that Alexa will make in the future.
Amazon calls the little digital errands that Alexa does "skills." This can mean anything from reading the news to turning off a lamp in the living room. Amazon created some of the skills, but a growing number of them are created by third-party developers. In the Alexa app, there are now more than 100 third-party skills. Some are more useful than others; I don’t need Alexa to tell me dumb jokes or make weird sounds, but I do want her to know how to call me an Uber.
Amazon has also opened the platform to connected home device makers, who can also create skills through single-purpose apps, for example. Two days ago, Amazon launched an API that developers can use to instruct Alexa in how to control their products. So any smart lighting company could teach Alexa to understand and execute an "Alexa, dim the kitchen lights" voice command from the user. The API can also let Alexa control switches and smart plugs and adjust smart thermostats. It’s likely that control of home security systems will be added to that list as well.
Amazon began beta testing the API last summer with some very notable partners, including Nest, Ecobee, Sensi, Samsung SmartThings, and Wink. As more appliance makers participate, the more useful Alexa becomes.
Another way Amazon is spreading Alexa to more third-party devices is through Alexa Voice Service (AVS). The service enables the maker of any kind of connected device to put the Alexa brain inside their product, provided that it has a microphone and speaker. So some connected alarm clock, for example, might suddenly gain the ability to play music, answer questions, and control smart home products.
There are many companies out there vying to provide the strongest connected home platform. Count Apple, Samsung (SmartThings), Nest, Google (with its Brillo OS), and many others among them. The eventual winner will be the one that works with most of the galaxy of connected home devices out there. Amazon may win that distinction, because it immediately hooks them up with a speaking and understanding intelligence in the home.
The foundation of Alexa is voice. Many of the smartest people in consumer technology readily profess that talking to our technology is a far more natural interaction than tapping buttons on a screen. Amazon spent a lot of time and resources on developing Alexa’s ability to clearly hear and comprehend what’s being said in the room, and also understand something about the context in which it’s being said. In short, Amazon is teaching Alexa to understand not only a user's words but, more importantly, their intent.
This is a both an audio engineering challenge and a machine learning challenge. The Echo device, for example, has a ring of six microphones situated around its top. When Alexa understands which of the microphones is closest to the source of my voice, the other microphones step aside to let the target mic listen. That’s part of the way Alexa can hear your voice even over loud music playing from her own speaker.
While Apple’s Siri and Google Now are tied to a phone (for the most part), Alexa is a free-standing home appliance that can hear commands and instructions from across the room or even in other rooms. This is perhaps the main reason that Alexa has, almost by accident, begun to look like a sensible choice for the de facto brain of the connected home.
The truth is, for many of us, the smartphone hasn’t lived up to its billing as the central personal computing system in our lives. We may need different helper devices and different bits of software for different settings and different tasks in our lives.
Last week at its Build conference, Microsoft introduced its new focus on bots—not apps—as the best way to harness information in the 21st century. We may need specialized bots to collect, manage, and even analyze the information we need to make decisions.
Microsoft’s "bot" concept correlates roughly to Amazon’s "skill" concept. Both technologies are in their infancy at this point. Microsoft is just now providing developers with the intelligence to build bots, while Amazon only recently began inviting third parties to create skills.
Microsoft also announced its own voice platform at Build. Like Alexa, it allows developers to quickly build voice and personal assistant features into their apps and devices.
Alexa also isn’t trapped inside garden walls, as other personal assistants are. The company’s integration agreements with home products like Nest, and its decision to offer the Alexa voice platform for use by other developers, signals a willingness to cross platform lines in order to increase the utility of Alexa. But it also signals a focus on extensibility, meaning that Alexa will grow smarter and smarter as more and more third-party products (and their data sets) connect to her.
In the movie, the OS falls in love with the movie’s main character (played by Joaquin Phoenix), and he falls in love with her, but she begins to drift away from him as she encounters more and larger intelligences in cyberspace. So never fall in love with an OS. (Amazon may one day ship the product with a warning sticker reading "Do not fall in love with Alexa.")
Actually, if you’ll indulge me for a moment, my own aspiration for Alexa is that she ceases to be a "device" and becomes built into the environment around it, whether that be the walls of a home or the dashboard of an automobile or the fabric of a pantsuit. At that point, and Alexa might get there, we may have to change her name to "Hal."