I want to know how can I directly return an utterance or a custom action (instead of having an intent or a text that is mapped to an intent) when a button is chosen. I need it because I’m developing a chatbot that has several button-based menus which in turn give access to other menus or have to execute actions without them having an associated intent (I need the match to be deterministic in those cases, without dependence on the intent classifier).
E.g. having in
domain.md the following (it’s a silly example, but it illustrates what I need):
utter_choose_department: - text: "Which department do you want to access?" buttons: - title: "Sales" payload: "/utter_sales_access" - title: "Marketing" payload: "/utter_mkting_access" - title: "HR" payload: "/utter_hr_access" ... utter_sales_access: - text: "Choose an option, please:" buttons: - title: "Contact sales" payload: "You can contact sales department at email@example.com" - title: "Schedule an appointment" payload: "/action_sales_appointment"