As humans, there are key aspects to our social interactions that we simply overlook or are downright ignorant of. For example, almost every conversation we have with each other almost guarantees a back and forth flow of dialogue— every statement or question is followed up by yet another set of statements of questions until the conversation simply fizzles out. What if this wasn’t the norm, however? What if conversations never entailed follow-up remarks to a central idea? Well then you might as well be talking to a wall at that point.
Unlike conversation between humans, when building a CUI(Conversational UI), these follow-up statements in a conversation don’t come naturally. The developer has to consciously program the chatbot to be able to handle all of these follow-up statements.
Now, this might seem obvious, but it’s worth pointing out that there is an infinite set of follow-up questions and statements that could be made in a conversation.
For example, if I want to get a smog check done on my car, there are a series of of relevant questions I could ask:
- What dates and times are available?
- What is the price range?
- Do I need to make an appointment in advance?
This is where the developer comes in. As developers, it is our responsibility to forsee all possible follow-up statements and create intents to handle them. To do this, we use the Follow-up Intent feature on Dialogflow.
Before we can handle the extra requests and follow-ups, we first need to have our main intent. This intent will handle requests for a smog check. We can tag “Smog Check” as an entity referring to “@maintenance-type”.
Now it is time to add a follow-up intent. Think of this as any intent that is relevant to the main intent.
Dialogflow offers basic pre-built intents, but we also have the option to create a custom follow-up intent and populate it with our own training phrases.
In this case, we want to ask for a preferred time and date, so let us create a custom intent.
Now let’s add a confirmation follow-up intent using the template”yes” intent provided by Dialogflow.
Notice how this intent is already populated with training phrases. These prebuilt templates are really convenient in this way, but they aren’t always applicable.
Congrats! You have just created a primitive conversational structure with follow-up intents in under five minutes. Play around with these follow-up intents and let me know what you come up with below.