In this documentation you find information on how to configure your RASA agent, including step-by-step instructions for setting up the required settings.
Requirements for setup:
- a running RASA server
- RASA server is started with the command line option
Settings on RASA platform
Please note: For the VoiceBot integration it is essential to set some intents in RASA.
The next part explains how these should be defined.
In the VoiceBot block, a start intent must be defined which is called “tenios_start”.
rules: - rule: Tenios start rule steps: - intent: tenios_start - action: utter_start
responses: utter_start: - text: "Hello, how can I help you?"
For your bot to function correctly, you should most likely define a fallback mechanism in RASA. There are multiple ways to achieve this, but one of the simplest ones might be to define an intent that is called “nlu_fallback” (see Fallback and Human Handoff )
rules: - rule: Ask the user to rephrase whenever they send a message with low NLU confidence steps: - intent: nlu_fallback - action: utter_please_rephrase
responses: utter_please_rephrase: - text: I'm sorry, I didn't quite understand that. Could you rephrase?
If you do not specify any fallback mechanism, then RASA will return an empty response. The VoiceBot block then interprets the empty response as end of conversation, and exits the block.
End of conversation
In order to end a conversation and exit the VoiceBot block, it is necessary to define an endpoint. Therefor you can either make RASA return an empty response text or you can use a custom output payload:
responses: utter_goodbye: - text: "Have a nice day" custom: end_of_conversation: true
The advantage of using a custom payload is, that in this case, the output text “Have a nice day” will be played before exiting the VoiceBot block.
Action mappings (optional)
If you want our Number Routing to perform certain actions (e.g. forwarding to a number, playing an announcement, …), depending on your RASA bot, you can achieve this with output mappings.
responses: utter_support_rep: - text: "One moment please, we will forward you to our support staff" custom: output_action: "fwd_support"
Once the RASA bot detects the intent, the audio (“One moment please […]) is played. Afterwards the call is transferred to the support staff.