Voice Interfaces and Contextual Intelligence

Here’s a scenario: You are leaving to meet your friend Mark and you want to tell Alexa this:

  • You: “Alexa, let Mark know when I am almost there”
  • Alexa: “Okay”

To solve this you need to answer:

  1. Which Mark is relevant now? and btw, is it Marc or Mark?
  2. Where is “there“?
  3. Where is Mark – or where is he supposed to be in the near future?
  4. Where am I?
  5. How fast am I approaching Mark? (to decide what almost means)
  6. How do I communicate with Mark? Text? Call? (or the brand new “echo calling”)?

Then a system that executes this as a delayed one-time action and automatically shares my location on his Google Map (till we meet).

Scenario 2: You don’t want to come home to a dark house.

  • You: “Alexa, turn the kitchen lights on when I come home at night”
  • Alexa: “Okay”

To solve you need:

  1. Where is home?
  2. Where am I?
  3. Is it dark (when I am almost home)? and is that what I mean when I say “at night“?
  4. Am I going home?
  5. How do I communicate with the system that controls the kitchen lights?
  6. When do I turn them on? (not a big deal for kitchen lights but could be for slow to brighten outdoor lights or the HVAC)
  7. Save and execute as a persistent action (“when” is actually whenever).

Scenario 3: You are cutting a not-yet-fully-thawed full chicken for dinner and the in-laws have arrived!

  • You: “Alexa, unlock the side door”
  • Alexa: “You ordered pizza for dinner earlier this week. Can you tell me where you ordered from?”
  • You: “Enzo’s”
  • Alexa: “Side door is unlocked”

All of this is contextual intelligence and it is essential for natural language processing to sound natural. Check our concept integration as an Alexa skill: idid-inc.com/my-avatar