I may be trying to do things incorrectly here, so please forgive me, as it’s my first attempt at creating a bot.
I have a workflow in mind (think: flowchart) that isn’t well-served by forms, as it’s really a sequence of detected intents and not real “information” the user is submitting that needs to be used later.
That being said, I’m also trying to further guardrail the user along certain paths. I’m attempting to do this with buttons. However, this involves creating intents that are essentially empty because the button is directly specifying the intent. And, for single-example intents, training generates a warning (which is frustrating).
For example, think of a story that results in the bot uttering “Do you like turtles?” with the button “yes” and the button “no.”
As of right now, the only thing I can see to do is to create an intent:
- intent: likes_turtles
examples: |
- like_turtles
And a button:
utter_likes turtles_question:
- text: "Do you like turtles?"
buttons:
- title: "Yes"
payload: '/likes_turtles'
- title: "No"
payload: '/dislikes_turtles'
This generates a training warning but does work.
Is this the right/best way?