There has been an exponential increase in the number of smart devices in the world around us. It seems that everything, from indoor lights to the front doorbell, allows you to control them remotely. This means that not only can you unlock the door to your apartment if the courier delivery person lands up while you’re in office, but you can also set it so that it senses your presence and opens automatically. This is the allure of connected devices—their ability to magically reduce friction in tiny ways that make life that much more comfortable.

As we accumulate more connected “stuff", there is an incentive to try to make them all work together. Wouldn’t it be wonderful if the coffee maker turned itself on as I switch my alarm off so that, by the time I get to the kitchen, a steaming cup of coffee is already waiting for me. Or if my house sensed that I am pulling into the garage and turns on the lights, making sure the air-conditioning is at the correct temperature and starts playing the same music I was listening to in the car just as I enter.

A number of tech companies are building home automation toolkits that serve as a hub for our smart devices, offering a centralized place from which they can be controlled. Today, these home automation hubs can get our connected devices to act in concert with each other so that we can, with just the press of a button, set up our rooms to achieve a desired mood. But this is just simple automation. The true leap will come when we can add intelligence to the mix.

With the recent explosion of smart speakers in the market, it seems as if that future is finally here. These speakers are thinly veiled AI (artificial intelligence) devices, equipped with advanced voice- recognition capabilities that allow them to accurately identify spoken commands. At present we use them to entertain us—to play music that we request and to answer our trivia questions. But they are capable of far, far more.

The obvious application would be to combine a smart speaker with a connected system so that we can control our homes with spoken commands. While this seems to be yet another way to smooth the friction in our daily lives, it has the less immediately apparent consequence of giving the intelligence that powers the speakers access to a number of other sources of information about us. By plugging directly into our smart doorbell, an intelligent machine will be able to learn who we let into the house and who we do not. By connecting to our smart refrigerator it will learn what we consume most frequently and the brands that we prefer. By accessing the various recipes stored in our home automation system it will know when the lights are usually switched on and will be able to infer what time we return home from work.

This clearly raises privacy concerns. But if we are talking privacy, what is somewhat more troubling is the speaker itself. Smart speakers have to remain on all the time if they are to respond when we command them. As a result, their voice-recognition engine is always listening to every word—understanding what is meant in the discussions that it overhears. An intelligent system will infer from these conversations, information about us that, when overlaid on top of all the other data accumulated from other connected appliances in our homes, will generate an incredibly accurate profile of us, our likes, dislikes and our personal preferences.

But what could anyone want with all this information?

The primary goal of e-commerce is to reduce the friction of the shopping experience. Today this is achieved by building recommendation engines that offer suggestions as to what we should buy next so that we don’t have to browse the catalogue before making our next purchase. However, at present, the accuracy of these prediction engines isn’t good enough for all their recommendations to be useful all the time.

This will inevitably improve—potentially to the point where e-commerce companies will be able to ship us products that they are certain we will want to buy—even before we ask for them. This will fundamentally transform e-commerce from a shop-then-ship to ship-then-shop model. It will allow e-commerce companies to shift their focus from marketing their inventory to building extensive distribution channels through which they can pre-emptively ship products, so confident in the accuracy of their prediction models that the low rate of returns will make this anticipatory shipping model profitable.

This is the long-term vision for the e-commerce ecosystem and the goal towards which everyone in the value chain is working. In order to deliver on this vision, shopping algorithms need deep knowledge about us—knowledge that they will only be able to obtain by observing us all the time, filing away our smallest reactions and behaviours to build as accurate a map of our preferences as possible.

A connected, intelligent home network is ideally suited to generate this level of insight. By connecting a network of sensors to an always-on voice-recognition engine and vertically integrating it into an intelligent system designed to smooth out the friction in our lives, we are paving the way for this utopian future. Many of us will delight at the wonder of a world where we don’t have to worry about buying milk or replenishing the bin liners before they run out. But at least some of us will cringe at how intrusive all of this will be.

Rahul Matthan is a partner at Trilegal. Ex Machina is a column on technology, law and everything in between. His Twitter handle is @matthan.

Comments are welcome at views@livemint.com

Close