HOW TO PREPARE YOUR ALEXA SKILLS FOR GOOGLE HOME
Google just announced, yesterday, a change of mind: they are going to release a SDK to let you add “agents,” the equivalent of Alexa skills, to Google Home, in December. If you are an Alexa developer, the first question that should come up is, “How can I port my skills over to Google Home?” Since they haven’t released the details of their interface, there is no way to know the exact answer to that right now. But there are certain steps you can take, per-emptively, to prepare for it.
In the Sub War example, produced by TsaTsaTzu, a layered architecture was given as best practice for isolating the different layers of our application from change. At the lowest level was the business logic. Keeping that separate from the interface logic means that you can build an audio app on it, and later add a traditional mobile or web app. Since the business logic is agnostic to the interface, it can be re-used and consistency maintained across instances.
In addition to separating out the audio logic into its own layer, the example also split out an Alexa layer. All of the Alexa specifics are dealt with in this layer. Amazon specific structures and objects are converted to neutral ones. The Audio Interface layer is written entirely in terms of those neutral objects.
This technique means that, when the Google Home API is released, all that needs be done is to write an interface layer for that API. (Assuming it isn’t wildly different.) Your audio logic and business logic remain unchanged. You can leverage your investment in your skill to a new platform, and provide a comparable experience there with the least amount of work.
When Google’s SDK is released, TsaTsaTzu (or someone else, since the project is public domain) will port SubWar over as a specific example. In the meantime, all the code is posted to GitHub along with thousands of words of documentation and nearly 50 minutes of video:
