Alexa Live notes

My notes about some sessions of Alexa Live 2019 conference. Full conference info here

• How to shift from screen-first to voice-first design

  • Principles
    • Be adaptable
      • Use lots of utterances and variations
      • Slots and synonymous
    • Be personal
      • Remember user information
      • Context & Situation
      • Think on retention
    • Be relatable
      • Be concise and straight to the point
      • In-skill purchases
    • Be available
      • Voice is everywhere
      • Take the initiative & lead the conversation
      • Don't try to convert a flow chart experience, categories and so on to a voice-fist experience
  • Design artefacts in a canvas - Storyboard (Adobe XD and Amazon templates)
    • Happy path
    • What happens when an user return to your skill
    • Alexa should remember previous information
    • Living documentation

• Tailor your Alexa skill responses to deliver truly personal experiences (dev blog post)

  The goal is to hide the technology behind the experience and being natural

  • Personal Experience
    • Answer any question only once
    • Preferences should become defaults
    • Context is kept thought sessions
    • Language should feel natural
    • Users feels their experience is unique
  • How would you create a personal experience in Alexa?
    • Memory
      • Session Attributes
        • Example: Different error messages depending on an error counter
      • Persistent Attributes
      • User information
        • Time Zones & Measurements
        • Customer Profile API
        • Account Linking
    • Variety
      • Randomization
      • Contextualization
      • Language adaptation
      • Localization & I18n
    • Localization

• How to build your first great Alexa skill

  • Technologies provided by Alexa
    • Speech recognition
    • Machine Learning
    • Natural Language Understanding
    • Text to Speech
  • Understanding the SDK
    • Handlers
  • Turn-by-turn conversation
    • Ask Follow-up questions
    • Auto-delegation
  • Memory
  • Amazon API > Settings API

• Alexa Skill Developer Tools: Build better skills faster

  • Design
    • Core values
      • Unconstrained user input
      • Sequential presentation
      • Ephemeral data
      • Text input is easy
    • Principles
      • Be adaptable
        • In grammar vs out of grammar
      • Be contextual
        • Be careful with random answers
        • Session attributes
      • Be available
        • Wide top level UI, not nested menus
  • Build
    • SMAPI
    • ASK CLI
    • VS Code integration
    • AWS CodeStar
  • Test
    • Test events for Lambda code using JSON from simulator
    • Beta testing
      • Pulselabs
  • Launch
    • Analytics
      • Interaction Path

• Best practices for testing your Alexa Skills (dev blog post)

  • Bespoken

  All code is guilty until proven innocent

  • Automation FTW
  • Types of testing
    • Unit testing
    • E2E
    • Continuous testing & monitoring
    • Usability performance testing

• How to write great dialogs for Alexa Skills (dev blog post)

  • Situational design
  • Brevity is King
  • Storyboard (Amazon templates for Adobe XD)
    • Start with your happy path
  • Be adaptable
  • Be brief
  • Handle confirmations
    • Implicit vs Explicit
  • Avoid voice menus
  • Error recovery > try to prevent errors in the first place!
  • Contextual errors
  • When to say sorry
  • Voice quality
    • SSML, punctuations & grammar rules
    • Split personality