Bringing sophisticated AI Technology To The Masses
At A Glance
Bringing artificial intelligence technology to the masses is no easy task. It takes a ton of computing and processing power, which is why it’s difficult to equip personal devices like smartphones with AI capabilities. Xnor.ai, a Seattle start-up spun off of the Allen Institute for AI, established commercial proof that they could enable sophisticated AI on low-cost devices at scale. Along its journey, Xnor partnered with GenUI to bring its vision to life.
App to enable AI on home monitoring device
One of Xnor’s earliest partners was Wyze, whose WyzeCam home security user base wanted fewer push notifications of motion detection, specifically only those related to person detection. In other words, only alert camera owners when a person enters their home, not something like a fly.
The first step to achieving that objective was to build Xnor’s AI technology into a mobile app that allowed for object detection of images, separating people from other objects. Xnor’s machine-learning engineers collaborated with GenUI’s iOS engineers to prototype an app on the iPhone, which was published to the AppStore and later integrated into WyzeCam’s mid-year 2019 release to over 1.2 million users. Person detection is now one of WyzeCam’s most appreciated features.
App to tell the Xnor.ai story
Xnor then asked GenUI to help prepare them for the all-important Consumer Electronics Show (CES) just a couple of weeks away by building an app to effectively tell the Xnor story.
GenUI helped to unify random apps, fix bugs, improve collaborative storytelling, curate experiences and visibly demonstrate how Xnor is overcoming the difficult challenges in deploying edge-based AI computing -- particularly as compared to cloud computing.
SDK for Developers to apply SI to Edge Technology
Lastly, Xnor turned to GenUI to build a self-service software development kit (SDK) by way of a Portal (dubbed AI2GO) for developers interested in applying AI to edge technology.
The Portal offered pre-tuned models based on speed and memory footprint across a number of hardware targets; a benchmarking tool that helped developers gauge the performance of their models; and a set of samples (counters, visualizers, etc.) available in Python, C and Swift.
“GenUI was an absolute joy to work with. From kick-off to hand-off they brought a level of creativity, attention to detail, and communication that allowed us to not only meet our deadlines but celebrate them as incredible successes. I can’t recommend GenUI and their talented team highly enough.”
Sophie Lebrecht, SVP Strategy and Operations, Xnor.ai