Mycroft Mark II

Mycroft AI Launches Privacy-Centered Mark II Smart Display

Open-source voice assistant developer Mycroft AI has launched the Mark II smart display after several years of development following a successful Kickstarter crowdfunding campaign. The customizable and privacy-centric smart display can serve as a smart home hub and perform much of the standard functions of an Amazon Echo or Google Home while also offering offline operations while not collecting any personal data.

Mycroft Mark II

The Mark II looks like a standard smart display, though with hints of a classic Apple II in its casing, though more than a third of the case’s plastic is recycled. The device comes with a far-field microphone array, a five-megapixel camera with infrared sensors, and a 4.3-inch color touchscreen. While capable of performing much of its tasks without the internet, the Mark II includes Bluetooth, Wi-Fi, and even Ethernet connectivity options. The custom operating system uniting the device is called Dinkum, as in the Australian phrase “fair dinkum,” referring to something honest and without deception. For smart homes, Mycroft is compatible with the Home Assistant operating system, which works with the vast majority of standard smart lights, switches, and other devices.

The smart display is far more modular than anything from Amazon or Google, however. The Mark II was designed to accommodate those who enjoy customizing their tech. There are four USB ports, two each for USB 2 and USB 3 for any hardware connection, media storage, or software transfer. Owners can also take apart and reassemble the Mark II with a screwdriver, augmenting and repairing it as needed. It’s come a long way from the initial prototypes four years ago, buoyed by enthusiastic backers and designers, a category with some crossover.

“I got into this because I was a Kickstarter backer. “I was a customer of Mycroft before I was the CEO. I’m very passionate about it,” Mycroft AI CEO Michael Lewis told Voicebot in an interview. “Dinkum is the best version of our software and the most private. It strips out profiling, so we don’t even have the capability to take user data since that’s just a liability if there’s a data leak or data breach. We’re not doing anything with, so why have it?”

Though Mycroft Mark II’s software comes ready to use, those interested in playing with the software can do so by picking a different flavor of Dinkum. There’s a downloadable version that runs on Raspberry Pi that’s largely the same as the one built into the Mark II but doesn’t include an automatic update mechanism. That way, the updates don’t mess with the customization and continue to keep the user’s data private.

“We’ve always had a policy of maximally protecting people’s privacy, but our privacy policy was put in place by a legal team that didn’t understand it, and it didn’t reflect how we actually operate,” Lewis said. We never actually collected data on people at all unless they opted in, and that data we didn’t sell. The only thing we ever thought about using it for was improving our speech-to-text models.”

The initial Kickstarted envisioned a relatively low-cost smart device, but that’s no longer the case. Lewis explained that the approximately $150 price point was based on a BOM (bill of materials) calculated at $99. Supply chain troubles, chip shortages, and other market woes drove the BOM up to nearly $300, necessitating a jump in price to $499, though those who pre-ordered were able to buy it for $349. With the official release, Mycroft has plans to iterate and upgrade the smart display with the support of its very involved user base.

“We will have more services in the future. I’m particularly excited for music streaming. I’d love to be able to provide a streaming service,” Lewis said. “But, we want to make sure that anything we officially support complies with our privacy rules and has a proper QA mechanism.”

Will Smart Displays Be Popular? Mycroft Mark II May Offer Insight.

Mycroft AI’s Legal War Against a ‘Patent Troll’ Heats Up

Mycroft AI Launches Mimic 3 Neural Text-to-Speech Platform