Intent-driven dialogs

To let users accomplish tasks with text or voice commands or get a response from the AI agent, you can add intents to the dialog script.

In Alan AI, intent is a function that allows it to identify a request or task behind the user’s utterance and take the appropriate action. For example, when building an AI agent for flight booking, you may want to include the following intents:

  • Search for available flights

  • Book an flight

  • Reserve a seat

Intent-driven dialogs are best suited when the user’s goal is well-defined, the dialog is structured and follows a rigid path. A typical dialog script will combine both tools and functions for intent-driven and intentless dialogs, such as question answering.

You can use the following Alan AI tools to design intent-driven dialogs:

User commands

Learn how to match the user’s request or task to intents and provide responses or take actions

Patterns

Learn how to define request and response phrases

Slots

Learn how to pick out important information from the user’s utterances

Contexts

Learn how to design a multi-branched conversation flow

Predefined objects

Learn what predefined objects Alan AI exposes

User data

Learn how to make a dialog more specific with user data

Error handling and re-prompts

Learn how to gracefully handle errors in the dialog

Lifecycle callbacks

Learn how to use callbacks as the dialog transitions between states