[yourapostasy]

Companies see the Cambrian explosions of growth (and equities appreciation) that follow successful UI shifts and want to reproduce that with voice UI.

You could do a hell of a lot by establishing a framework for third parties to use such that they don’t stomp upon each other and the system-reserved words/phrases, as well as homophone rejection/discrimination, then brute-forcing lots of recognized cases. Right now, I’m seeing second-system effects in voice UI efforts by everyone (possibly excepting the dedicated voice recognition outfits like Nuance), by trying to apply a general machine learning approach to all of it and eschewing brute-force as not pure/clean enough.

[jxdxbx]

Well, Alexa seems to be going the brute force route--using third-party skills is just like using a spoken command line utility. The problem is, while this approach is simple and it works, very few skills are useful.