Intents are one of the key pillars in conversational AI applications.  They represent an action a user wants to carry out and is expressed via natural language in the utterance.

Intents can be thought of as a goal the user wants to reach, such as placing an order, paying a bill, or adding an item to a list.

When an utterance is supplied to a conversational AI application, the natural language processing (NLP) capability will attempt to match the incoming utterance to an intent using any training utterances to try and capture the user’s underlying intent.

A sufficient number of utterances must be suppled for each intent for it to detect the different types of utterances that users may supply. We recommend 10 to 15 to start off with.

You can see a simple Welcome intent with a list of sample user utterances:

# Welcome
-Hi
-Hello
-How are you?
-Greetings!
-Hi friend!
-Hello there
-How are you today?
-How can I help you today?
-What can I help you with?
-Hello friend!

Next

Learn about the concept of slot filling and the relationship it has to your conversational AI application.