In December, Google opened up Google Assistant for developers with their Actions on the Google platform — and as with any emerging technology, we here at TribalScale were eager to begin working with it. It is clear that Google’s new voice assistant is meant to be a direct competitor to Amazon’s popular and more established Alexa. Having already worked extensively on the Amazon Alexa platform, our engineering team was interested to see how the Assistant on the Google Home device compared.
While Alexa has Skills — applications which can be installed through the Alexa Skills Store — Google Home has what is calls Actions. Actions differ from Skills in that there is no need to install anything, the user can just prompt their device by saying “OK Google” followed by the name of the service it needs. The prompt “OK Google, Talk to Personal Chef” will immediately start the built in Personal Chef Action. Google hopes that this will make Actions more conversational and more discoverable. This also removes a barrier in obtaining users, in that there is no extra step of installing or enabling an app/skill. All Actions will be available on Google Home ‘out of the box’, with more and more actions becoming available for you to use without you doing anything.
There are two types of Actions: Direct Actions and Conversational Actions. Direct Actions are similar to how most interactions with other voice assistants work, like Apple’s Siri– you ask a question or make a command (“Play a song by The Beatles”), you get an answer. Conversational Actions on the other hand involve more back-and-forth between the Assistant and the user. Google’s acquisition of API.AI last September has also added the ability for developers to easily utilize API.AI’s natural language processing capabilities for their Actions to simulate a more natural person-to-person conversation. Rather than giving the Assistant a very specific command, the user is able to give it more vague input and still get a response. For example, “I need an Uber,” “Call me an Uber,” “Order an Uber” and “Start Uber” will all allow you to call an Uber. API.AI also allows developers to anticipate this more vague input throughout the conversation flow, giving the user the ability to express the same intent in a variety of ways.
Evolution of Google Home & other VUIs
While Google Home and it’s Assistant still have a lot of ground to cover in terms of available functionality (Amazon Alexa currently has over 5,000 skills available in it’s Skills Store), the Assistant is available on a number of platforms besides the Home. It is available on Google’s Pixel phones and Google Allo, allowing Actions that were developed for Google Home to also become available on millions of mobile devices. Just last month, Google announced that they’ll be bringing the Assistant to Android TV, and at CES 2017 we saw that both the Assistant and Alexa will soon be brought to vehicles as well. It will be exciting to see how Google’s Assistant and Google Home evolve, as we continue to see voice user interfaces continue to become a bigger force.
Chloe is a Lead Engineer at TribalScale. In her free time she enjoys travelling, craft beer, yoga, and obsessing about her cat.