Alexa Bristol Workshop November 2017

Attended an Alexa Workshop on the 3rd November 2017 in Bristol hosted by Amazon, these are my notes.

Introduction

Hosted by David Low

Twitter Accounts

Alexa Design Alexa Build

Course sign-up link

Amazon Developer Amazon AWS

Theory

Mobile moment: when someone picks up a device to achieve a goal. e.g. watching football in front of the tv, get extra detail. Ask person next to you, why not shout it towards the tv.

Disruptions and Expectations

Expectation is shifting

70s. Character mode => Teletext 80s. Graphical Interaction 90s. Web 00s. Mobile Smartphone Present. Voice Interface

Act less like computers and more like people.

Why Voice

Walt Mossberg: ambient computing, the transformation of the environment… photo taken

The Alexa Service

  1. Alexa Skills Kit
    • SDK can connect to anything with linux, a microphone and speaker.
  2. Amazon Alexa
    • Lives in the cloud.
    • Automated Speech Recognition.
    • Natural Language Understanding.
    • Always learning.
  3. Alexa Voice Services
    • missed slide.

DVLA => Not losing jobs, taking away the short, repetitive tasks (easy answers), focus on the harder, longer ones.

Under the hood

  1. User says something (mp3 file to the cloud).
  2. Apply Natural Language Model to figure out the intent.
  3. The intent gets sent to the back end.
  4. Your service then processes the request.
  5. Respond to intent through text and visual.
  6. You pass back a textual or audio response / You pass back a graphical response.

Two Ends of a Skill

Alexa at heart is a voice service and will always be a voice service. Weather: assumes the request is for now, rather than any other time.

Example Skill - Coffee Now

Invocation words and names

Open, launch and play are all the same. Ask, tell

Alexa, ask National Rail for my commute.
Alexa, open Just Eat.
Alexa, tell Uber to get me a ride.
Alexa, launch Cat Facts.
Alexa, play Reindeer Trivia.
Alexa, ask Coffee Now.

Intents and Slots

Sample Utterances

Photo taken.

Custom Slots

Photo taken.

Putting It All Together

Alexa Always start here ask Invocation word Coffee Now Invocation name to order me Intent: OrderCoffee a large Slot: CoffeeSize skinny Slot: CoffeeMilk cappuccino Slot: CoffeeType please

Types of Request


Morning Coffee Break


CODING DEMO


LUNCH


Developer Advocate - Talk from Jamie Grossman

Intents

Slots

Utterances

When building skills, focus on your interaction models.

To test apps online you can use Echosim