I had the idea for the wand in lockdown, 2020, just after my first son was born. Two things were on my mind: I didn't want my kids growing up staring at screens, and I was watching my elderly parents struggle for dignity thumb-typing WhatsApp. A camera in the living room was never an option — I'm from the audio industry; privacy had to be physics, not policy.
I’ve spent 20 years on edge DSP and edge AI: earphones, the stereo-to-surround upmixer that went into Dolby Atmos, MPEG-I immersive-audio standards, and most recently neural networks running inside earbuds. The wand is the same discipline pointed at motion instead of sound.
What it is: a handheld device that types what you write in the air. A small neural network runs on the device itself — Nordic silicon, under 30 ms per character — and it connects as a standard Bluetooth keyboard, so it works with TVs, phones and computers with nothing to install. Today’s device has no microphone and no camera. It learns your handwriting.
First market is the TV, the screen where typing is worst. But the reason I’m building this is bigger: computing is heading away from screens, and the interface that replaces them is a hand and a voice — gesture out, push-to-talk speech to a language model in, everything processed on the device, nothing leaving your hand unless you send it. Screen-free computing needs an instrument. That’s what we’re building.
Try the recogniser right now — your phone becomes the wand: phone-demo.the-wand.ai/write
— John Usher, co-founder, Input-A (Berlin)