Amazon’s Alexa engineers are always working behind the scenes to add new features and functionality not just for consumers, but developers as well. Today’s post offers an Amazon Alexa Developer Blog roundup, to point the way to some timely new blog posts Alexa skill devs are sure to find informative.
From Echo Buttons To HypRank, These Posts Will Get You Up To Speed
The first two are worth reading because they explain the thinking and engineering behind a recent change to the Alexa software that directs users toward one or more skills that may be able to handle their request when the native Alexa software can’t. Because your skills will be subject to this new AI ranking method, knowing how it works can help you take advantage of the method when designing your skills—or at least, prevent any negative impacts.
The Scalable Neural Architecture behind Alexa’s Ability to Select Skills
Four out of five Alexa customers with an Echo device have used a third-party skill, but Amazon’s always looking for ways to make it easier for customers to find and engage with skills. For example, Amazon recently announced they are moving toward skill invocation that doesn’t require mentioning a skill by name. Alexa uses a two-step, scalable, and efficient neural shortlisting-reranking approach to find the most relevant skill for a given utterance. This post describes the first of those two steps, which relies on a neural model we call Shortlister.
HypRank: How Alexa Determines What Skill Can Best Meet a Customer’s Need
The shortlisting step uses a scalable neural model to efficiently find the optimal (k-best) candidate skills for handling a particular utterance; the re-ranking step uses rich contextual signals to find the most relevant of those skills. We use the term “re-ranking” since we improve upon the initial confidence score provided by the shortlisting step.
This next one is one installment from Amazon’s Things Every Alexa Skill Should Do series. You can download the entire guide here.
Things Every Alexa Skill Should Do: Simplify Choices
Your skill might have the ability to handle user interactions with some well-crafted slots and utterances, but what about when they respond with “yes?” By making your phrasing more specific, you give customers more confidence in their answers, and also get the kinds of responses you want.
For those who’ve come to Alexa skill dev from a web dev or general software programming background, this next one’s a helpful read.
How Building for Voice Differs from Building for the Screen: Talk with Your Customers, Not at Them
Graphical user interfaces (GUIs) are inherently inflexible; customers have to first learn how to navigate them before they can use them to find what they need. A good GUI presents a consistent path each time so users can habitually perform the task and achieve the same goal every time in minimal time. It is declarative and inherently non-cooperative. A good voice-first UI is cooperative because conversations are cooperative.
Finally, devs who want to get started with a skill for Echo buttons will find a lot of good information and guidance in this last post.
Echo Buttons Skill Teardown: Don’t Cut That Wire
Don’t Cut That Wire is a pattern-matching game where Alexa shows the customer a sequence of colors on two or more Echo Buttons. The customer defuses a “bomb” by pressing the buttons in the same order. In this skill teardown, we uncover what makes this skill engaging for Alexa players and share tips that you can apply to your own Echo Buttons skill.