API.AI lets you build and deploy Conversation Actions directly within an easy-to-use, web GUI. This document goes over a few main concepts of API.AI that help you build actions.

See the rest of the API.AI documentation for more information on important concepts and how to use the tool.

Building Conversation Actions

The following general steps describe the process of building actions in API.AI from design to deployment:

  1. Design the dialogs, or user interface, for your action. Writing the dialogs upfront lets you catch mistakes, creates a natural sounding conversation, and gives you a good point of reference to code your actions when you're ready.
  2. Build actions with intents in API.AI, which let you define invocation and discovery, dialogs and fulfillment, and more.
  3. Test the action with the Google Home Web Simulator.
  4. Ensure you've gone through the launch checklist.
  5. Configure and deploy the action.

Learn more

See the end-to-end tutorials that go over building simple to complex Conversation Actions.

Getting Started
A very simple "hello world" type sample that asks a question to ascertain if you can be buddies with the action's persona. This sample gets you started with API.AI and has no dependencies on webhook fulfillment.
Facts about Google
A simple action that shows how to do webhook fulfillment and how to use contexts.
Guess a Number
A complex action with an involved dialog that shows how to use all of the major API.AI features.