CES 2017 was full of screens and gadgets as usual. Thinner televisions, faster laptops, concept cars with touchscreens in places touchscreens have no business being. But the thing everyone kept coming back to was a small cylinder that just listened.
The Amazon Echo had been out since 2014, but 2017 felt like the moment it stopped being a novelty and started being infrastructure. Almost every major manufacturer at the show had Alexa built into something. Refrigerators. Thermostats. Cars. Light switches. The running joke by day three was that it was harder to find a product without Alexa than with it.
What made this interesting was not the hardware partnerships. It was what they revealed about the direction Amazon was betting on. Voice was not a feature being added to products. It was becoming the interface that products were built around. The assumption was that if Alexa was everywhere, the value would compound. More devices meant more interactions. More interactions meant better understanding. Better understanding meant a more useful assistant.
Google had Google Home, which had just launched a few months earlier. Apple had Siri, but Siri was still mostly a phone thing. Amazon had gotten there first and had spent that time building a developer ecosystem around it. The number of Alexa skills had grown from a few hundred to several thousand. Most of them were not very good. But the platform was there.
I remember watching a demonstration where someone asked a refrigerator to add milk to their shopping list, and then confirmed the order on their phone. The friction involved in those extra steps made it slightly less impressive than intended. But the direction was clear. The goal was not to make refrigerators smarter. The goal was to make the gap between thinking you need something and having it ordered as small as possible.
The smart home had been promised for a long time. Automated thermostats and smart bulbs had existed for years. What was different in 2017 was that a common language was starting to emerge. Not a perfect one. Not a universal one. But enough of one that devices from different manufacturers could respond to the same voice, and that felt genuinely new.
The question nobody at CES was asking, and that would take a few more years to surface properly, was what it meant to have a permanently listening device in your home and who was responsible for the data it collected. That question was coming. It just had not arrived yet.