Mark Zuckerberg tried to connect his DIY smart home to a gray t-shirt cannon – The Verge

December 20, 2016 Facebook Twitter LinkedIn Google+ DIY Project

In a post today, Mark Zuckerberg revealed the details of his ambitious year-long project to build his own “Home AI.” Dubbed Jarvis, the system is basically an advanced Smart Home, letting you give voice and text commands to any device in the house. The Facebook CEO is a little late in the reveal — he initially said he’d demo the system in September — but the result is a fascinating look at exactly how far you can get home-brewing your own Smart Home.

Zuckerberg built the system in Python, PHP and Objective C — but like most programming problems, the bulk of the work seems to have been stitching together existing systems and codebases. That meant connecting a Crestron lighting and thermostat system with a Nest camera, a Spotify/Sonos speaker system and a Samsung TV. On the input side, he used established APIs to enable voice and mobile text commands.

Zuckerberg also experimented with physical devices, finding that even a simple device like a toaster is still fairly difficult to connect to a master control. “Connecting a food dispenser for Beast or a grey t-shirt cannon would require hardware modifications to work,” Zuckerberg wrote.

The final product seems like it’s still a few steps behind what you’d get from an Echo or Google Home. When Zuckerberg demoed the system for Fast Company, he had to tell the computer to turn the lights off four times before the command went through, and was suitably embarrassed. Aside from the t-shirt cannon, it also doesn’t do much you couldn’t do with consumer devices — but given that it was built by a single developer in his spare time, perhaps that isn’t so surprising.

Still, it’s not an entirely frivolous project for a CEO like Zuckerberg. Open-source AI software is a big part of what Facebook produces these days, part of the company’s larger efforts to encourage developers to use their tools and build on their platform. Building a home system let him test out those tools in practice, and see exactly how far the smart home ecosystem has to go before it can connect an entire home out of the box.