Having just released an update for a mobile app and getting Alexa up and running a few platforms, we’ve been learning first hand how the complexity of a system can expand exponentially as new features are offered. This is a challenge that will be facing Alexa Voice Service developers over the next few years.
There are a myriad (at least) of variables that affect an AVS implementation. First, there’s the hardware (application processor), then there’s the DSP, then there’s the wake up source, the OS/platform, and after that, ensuring that the API client can support all of the new features of AVS as they come out (Alexa Calling, notifications, music, cards, etc.). At some point, the company doing an AVS product might want to expand to other devices, which means having to re-test.
What might make this even more difficult is when Skills start to offer specialized services that might involve the hardware (an example would be sensor data collection). This is a challenge that Amazon (or Google… but probably not Apple) will face in the coming year or two — how to ensure support and testing across multiple deployments of Alexa on different hardware.