

I have added the swipe library to FUTO. It works, but not well enough for me. (That may be my clumsy swiping, but the issue remains.) I estimate that it gets about one word in four wrong. GBoard is still wrong about one time in ten, but at that rate, swiping and correcting is still much faster for me than traditional typing.
Voice recognition with keyword commands works quite well and reliably. AI-based voice assistants are often not even “better than nothing”. In the time it gets me to argue the AI into doing what I want, I could park my car and write a better non-AI menuing system, to say nothing of using one.