Google re-acquired its own Nest thermostat back from its parents company, Alphabet, earlier this year (it makes sense if you don’t overthink it), and already the innovations are rolling out. The latest change to the company’s smart-home product line? More intuitive software powering the security cameras, which stop alerting your phone every time the cat walks in the room, but instead can detect whether the creature it sees in the frame is human or not.


The new process relies on Google’s deep learning software to analyze the footage from the home’s security camera and then determine whether or not the image shows a human. Users will get one of two alerts: that’s probably a human in your living room, or that’s definitely a human in your living room. Eventually, Nest plans for its software to become so smart that it can tell you precisely which person is in your house, whether it’s your mom dropping off a casserole or the scary guy from up the street.

Now a little bad news for the tinfoil hat crowd…yes, in order for Nest to get this smart, it has to take your camera’s feed and send it over to its servers to be stored and studied. For reasons that make sense if you’re not squeamish about having your house recorded and sent to faceless servers somewhere, it has to store the footage in order to create a database of feeds for “intelligence gathering” analysis. Security cameras can’t get smarter if the software doesn’t know what it’s looking at, and what better way to make that happen than with lots and lots of practice? Nest, for its part, says this footage is stored on its own servers, not Google’s, so your not in danger of having hours and hours of your daily life sent to the search engine giant’s network.