A virtual assistant displays multiple answers from different states
If the virtual assistant in the try-out window displays information from multiple states at the beginning of the conversation without waiting for user input and responds to user input that does not match the expected intent, make sure that a condition is defined for the transition that connects the states - intent, entity, or a prerequisite.
If the transition with no conditions connects two states, it allows the virtual assistant to move freely from one state to the next automatically without stopping and without user's input evaluation.
User input is everything the user writes, including clicking the menu button. Therefore, if there is a need to guide the user through the guided dialogue with menus and buttons, a condition (most commonly, intent) must be specified for each transition between the states where buttons are defined.
To understand how the virtual assistant moves in the dialog, see the Conversations view for conversation analysis.