The question nobody was asking
Most accessible input tools I've looked at work the same way. They translate someone's gesture, or their eye movement, or their switch press, into something a second system then interprets. That's useful, and for a lot of people it's the only option that exists. But it treats the person's hands, or eyes, or whatever they're using, as a workaround. Not as a keyboard.
I kept coming back to a simpler question: why does it have to work that way?
A keyboard is just a way of turning a physical action into a letter on a screen. Press a key, a letter appears. There's no law of nature that says the physical action has to be a key press. If a hand gesture can be recognised reliably enough, it can be the input itself, typed straight into WhatsApp, Gmail, or anything else, exactly the way a keyboard would.
That distinction matters more than it sounds like it should. A translator needs somewhere to send its translation: an app that's been built to receive it, a developer who agreed to support it, a standard everyone involved has signed up to. A keyboard doesn't need any of that. Every app that exists already knows what to do with a keystroke. It was built for one, whether or not its developer ever thought about who'd be pressing the keys.
So instead of asking "how do I translate a gesture into something an app can use," I started asking "how do I make a gesture into the same signal a keyboard already sends." That's the whole idea behind GestureKey. Not a new interface someone else has to learn to support. Just a different way of producing the same keystrokes every app already accepts.
It's a small reframing, but it changes what you build. You stop designing a translator and start designing a keyboard. And a keyboard's job is simple: work for the person using it, and stay invisible to everything else.