Integrate Google Dialogflow CX

Before completing the tasks on this page, be sure that the prerequisites for the integration have been met.

Create a Script

Permissions Required: View Scripts, Create/Edit Scripts

You need to create at least one Studio script to route interactions to your virtual agent. Follow the script guidelines and requirements for integrating a virtual agent. The supported virtual agent action for this provider is VoiceBot Exchange.

• VoiceBot Exchange
• Voicebot Conversation

If you're using a SIP Protocol used for signaling and controlling multimedia communication sessions such as voice and video calls. backchannel connection with your virtual agent, your script has different requirements.

You can learn more about Studio scripting in the Fundamentals and Technical Reference Guide sections of the online help.

Custom Payload Best Practices

Follow these best practices when integrating custom payload data from Google Dialogflow CX into CXone: 

• Dialogflow CX doesn't use contexts to pass data to Dialogflow intents The meaning or purpose behind what a contact says/types; what the contact wants to communicate or accomplish like Dialogflow ES does.

• You can pass custom data to Dialogflow CX using JSON key-value pairs. In a Snippet action in your script, create a dynamic customPayload object and add the key-value pairs to it. For example:

  DYNAMIC customPayload
  customPayload.ani = ani
  customPayload.contactID = contactId
  customPayload.masterContactId = masterId
  customPayloadJSON = "{customPayload.asJSON()}"	

• In the Exchange or Conversation action in your script, configure the customPayload property with variable that has the asJSON() function in its value. You can find this variable in the customPayload object.
• Pass the customPayload JSON to the virtual agent using the Payload property of QueryParameters. See Google documentation on QueryParameters for Google Dialogflow CX.
• Data passed through QueryParameters is received by a webhook in Dialogflow CX. You can write code in the Dialogflow CX console to handle the passed data.
• Do not nest an object within the customPayload object. Nested objects are sent as literal strings.
• To pass custom data from your Dialogflow CX virtual agent back to the script, use the Custom Payload field in the Dialogflow CX console. Make sure you're in the console for the virtual agent you're using with CXone. Map this to your script using the customPayloadVarName (out) variable in the voice or chat Studio action in your script. For example, you can use this to set the next prompt's behaviors.

• Parameters set using customPayload can only be used in the external webhook. If you want to set parameters for use outside the external webhook, set them in a Snippet action with the session_params field. For example:

  {
  	"session_params":
  	{ 
  		"name": "Winnie Le Pooh"
  		"job": "Food critic"
  		"location": "Hundred Acre Wood"
  	}
  }

  Access the session parameters in the Dialogflow CX agent intent using the following syntax:

  $session.params.name = Winnie Le Pooh

  $session.params.job = Food critic

  $session.params.location = 100 Acre Wood

  Session parameters are only used with Dialogflow CX virtual agents. To achieve a similar result with Dialogflow ES, use contexts.

• Speech context hints can be passed with custom payload in the speechContexts parameter. The value of speechContexts.phrases must be a Google class token for the hint you want to give. The token must match the language and locale of your contacts. For example:

  DYNAMIC customPayload
  customPayload.speechContexts.phrases="$OOV_CLASS_ALPHANUMERIC_SEQUENCE"
  customPayload.speechContexts.boost=10		

• When using VOICEBOT EXCHANGE: To configure how long the virtual agent waits when the contact pauses while speaking, add the maxPostEnergySilenceMS parameter to the Next Prompt Behaviorsor Default Next Prompt Behaviors Snippet action.

Add an App to Virtual Agent Hub

Permissions Required: View Scripts, Create/Edit Scripts

Adding a configuration app to Virtual Agent Hub establishes the connection between CXone and your virtual agent provider.

If you're using a SIP Protocol used for signaling and controlling multimedia communication sessions such as voice and video calls. backchannel connection with your virtual agent, follow the steps for setting up this kind of connection.

1. In Studio, open or create a script and add a virtual agent action, if it doesn't contain one already.
2. Double-click the virtual agent action to open Virtual Agent Hub.
3. Click Add Bot.

4. Enter a Virtual Agent (Bot) Name and click Google Dialogflow CX in the list of virtual agent providers.

5. Click Next.

6. Complete the fields in the configuration form and click Next. You can find details about this form in the Configuration Page Details section on this page.

7. Click Next.

8. See the Configure Voice Integration Options section on this page.

9. Click Next.

10. On the Transcript page, select the option for whether you want the conversation transcript and intent information captured. You can choose to Capture and Archive Intent Info Only, to Capture and Archive Transcript and Intent Info, or to Capture and Archive Transcript Only. You can also choose Do Not Capture or Archive Transcript or Intent Info. This option is only available for utterance-based voice virtual agents. It's not supported for voice integrations that use an SIP Protocol used for signaling and controlling multimedia communication sessions such as voice and video calls. backchannel.

    Learn more about fields in this step

    Option	Details
    Do Not Capture or Archive Transcript or Intent Info	Select this option if you don't want to capture conversation transcriptions or intent The meaning or purpose behind what a contact says/types; what the contact wants to communicate or accomplish information.
    Capture and Archive Intent Info Only	Select this option if you want to want to capture intent information only. Captured data is stored only for the duration of the interaction by default. This option requires custom scripting.
    Capture and Archive Transcript and Intent Info	Select this option if you want to capture both transcript and intent data. Captured data is only stored for the duration of the interaction by default. This option requires custom scripting.

11. Click Next.

12. On the Test + Add page, you can test your bot's connection by sending voice or text prompts. If your test messages don't work, click Back to return to the Configuration page and verify the settings. You may also need to verify configurations in your virtual agent provider's management console.

13. Click Add.

Google Dialogflow CX Configuration Page Details

Use the information in the following table when completing the Configuration page. You see this page when adding an app in Virtual Agent Hub

If you're setting up Dialogflow CX to use a SIP Protocol used for signaling and controlling multimedia communication sessions such as voice and video calls. backchannel connection, follow the steps for setting up the connection instead of the steps in this section.

Configure Google Dialogflow CX TTS Output

Complete the steps in this section if you want to use the Google TTS service. The fields in the following image appear when you enable the Perform Output TTS option on the Dialogflow CX Configuration page in Virtual Agent Hub.

1. In Virtual Agent Hub, add a new configuration app and configure it according to the steps in Add an App to Virtual Agent Hub section on this page.
2. EnablePerform Output TTS.

3. To use a custom voice model, enable Use Custom Voice and enter the name of the AutoML model you want to use in the Custom Voice Model field.

   View image

   

   Learn more about the fields in this step

   Field	Details
   Use Custom Voice	Allows you to use an AutoML (automated machine learning) model to synthesize a custom voice for your bot. Dialogflow uses this voice instead of a TTS voice. If you use this option, enter the name of the AutoML model that you want to use in the Custom Voice Model field that appears. This field is only visible when Perform TTS Output is enabled.

4. If you want to use a Google voice model, select the TTS Voice Name from the drop-down.

5. Select the machine learning Model and Model Variant that you want the Google TTS services to use when synthesizing speech.

   Learn more about the fields in this step

   Field	Details
   Model	Select the machine learning model that you want the transcription service to use when transcribing the audio. These models have been trained to recognize speech from specific audio types and sources. Configure this field for voice virtual agents only. Not all languages support every model. Refer to the Google documentation for information about language support. Phone Call is the recommended option for most voice virtual agents. Available options are:  Command and Search: Use when transcribing shorter audio clips, such as commands or voice search. Consider using Latest Short instead. Default: Use when the audio doesn't fit the other models. It can transcribe any type of audio, but results may be lower quality for some types. Latest Long: Use with long-form content such as spontaneous speech or conversations. Latest Short: Use when the utterances are a few seconds long, such as commands. Consider using this model instead of Command and Search. Medical Conversation: Use when transcribing a conversation between a medical professional and a patient. Medical Dictation: Use when transcribing notes dictated by a medical professional. Phone Call: Use for transcribing the audio from a phone call. Video: Use when transcribing audio that has multiple speakers, such as from video clips or podcasts. If Video isn't available in the language you need, use Latest Long instead.
   Model Variant	Select the variant of the selected Model you want the virtual agent to use. Some models have an enhanced variant that provides higher quality results. Refer to the Dialogflow documentation for more information about the models that support an enhanced variant. Enhanced models are billed at a higher rate than standard models. For Phone Call models, the recommended variant is Enhanced. Available options for this field are:  Best Available: Select if you want Dialogflow to use the enhanced model if it's available for the selected model. If it's not, the standard model is used. Enhanced:  Select to use the enhanced model. If an enhanced model isn't available for the selected Model, Dialogflow uses the standard model. Standard:  Select to use the standard model even if an enhanced model is available. Unspecified: Select if you aren't sure which variant to use. Dialogflow defaults to the best available variant of the selected model.

6. To add silence to the start of the virtual agent's audio response, enter the number of milliseconds of silence to add in the Add Leading Silence field. The recommended amount of silence is 500 milliseconds.

7. Enter the Speaking Rate, Pitch, and Volume Gain to configure how you want the TTS voice to sound.

   Learn more about the fields in this step

   Field	Details
   Speaking Rate	Click + or – to change how fast the TTS voice speaks. The rate can be between 0.25 and 4.0. A normal speaking rate is 1.0. A rate of 2.0 is twice as fast as normal and 0.5 is half as fast as normal. The default is 0. This field is only visible when Perform TTS Output is enabled.
   Pitch	Click + or – to change the pitch of the TTS voice. Pitch is the highness or lowness of the voice. The default is 0. The maximum is 20.0 and means the voice is 20 semitones higher than the default. The minimum is -20.0 and means that the voice is 20 semitones lower than the default. This field is only visible when Perform TTS Output is On.
   Volume Gain	Click + or – to change the volume of the TTS voice measured in decibels (dB). The default is 0. The range is 16.0 to -96.0. A gain of +6.0 plays at approximately twice the volume of the default. Do not exceed +10 dB. This field is only visible when Perform TTS Output is On.

8. Return to the Add an App to Virtual Agent Hub section  on this page to finish adding the configuration app.

Configure Voice Integration Options

Permissions Required: View Scripts, Create/Edit Scripts

If you're using a SIP backchannel connection with your voice virtual agent, follow the steps for setting up SIP Backchannel instead.

1. In Virtual Agent Hub, add a new configuration app and configure it according to the steps in the Add an App to Virtual Agent Hub section  on this page.

2. Click Next on the Configuration page.

3. On the Voice page under Voice Integration Options, select Turn by Turn Control (CXone Voice Bot Gateway).

4. Under Text to Speech Options, select Use CXone Cloud TTS if you want to use the CXone TTS option. To use the virtual agent provider's TTS service instead, clear this checkbox and configure TTS output on the Configuration page of the wizard.

   If the form won't allow you to select Use CXone Cloud TTS, click Back to return to the Configuration page. Click the Perform Output TTS slider to Off. You can now enable Cloud TTS on the Voice page.

5. Select a TTS Vendor, if you're using Cloud TTS. If you're using your virtual agent provider's TTS service, this field is not visible and you can skip this step.

6. Under Transcription Options, select the option for the transcription service you want to use.

   Learn more about fields in this step

   Field

   Details

   Transcription Options

   Select the service you want to use for transcription: Use bot account transcription: Select this if your want to use your virtual agent provider's transcription service. Audio from the contact's utterances is sent to the provider before being transcribed. Use CXone transcription profile: Select this if you want to use a CXone transcription profile. You can learn more about this option on the CXone Turn-by-Turn Transcription help page.

7. If you choose to use a CXone transcription profile, click Transcription Profile to select the one you want to use. Click Add New Turn by Turn Transcription Profile to create a new profile.
8. Click Next.
9. Return to the Add an App to Virtual Agent Hub section  on this page to finish adding the configuration app.

Assign the Google Dialogflow CX App to an Action

Permissions Required: View Scripts, Create/Edit Scripts

When the virtual agent The meaning or purpose behind what a contact says/types; what the contact wants to communicate or accomplish app is set up and configured in Virtual Agent Hub, you need to assign it to the virtual agent actions in your script. This connects the action and the script to your virtual agent provider.

The action you use depends on your virtual agent and what it supports.

1. In Studio, open the script you want to modify.

2. If you haven't done so already, click the Tools tab, expand the Automation and AI category, and add the virtual agent action you want to use. Be sure to put it in the correct location in your script. If you don't see the action you expect, verify that your script media type is correct. For example, if you're creating a script for a voice virtual agent, the script must be a phone script.

3. Double-click the action to open Virtual Agent Hub.

   View image

   
4. In the left column, locate the virtual agent that you want to assign to the action and click the selection option next to it.
5. Click Close.

Change the Transcription Profile

Permissions Required: View Scripts, Create/Edit Scripts

A transcription profile allows you to use a different transcription service from the one the virtual agent provider offers. Each virtual agent can have one transcription profile. You only need to follow the steps in this section if you are working with a voice virtual agent and need to change the transcription profile.

1. In Studio, open or create a script and add a virtual agent action, if it doesn't contain one already.
2. Double-click the virtual agent action to open Virtual Agent Hub.
3. In the list of virtual agent configuration apps on the left side of the Virtual Agent Hub page, click the one you want to modify.
4. In the center of the virtual agent app properties page, click the Voice tab.

5. Under Transcription Options, select Use CXone transcription Profile if it isn't already selected.

6. Select the Turn by Turn Transcription Profile you want this virtual agent integration to use. If the option you want to use isn't there, you can add it.

7. Click Save.
8. Test your transcription service by using the microphone option in the Test section of your virtual agent app properties page.

Test Your Virtual Agent

Permissions Required: View Scripts, Create/Edit Scripts

You can test many aspects of your virtual agent to make sure everything works as intended. You can run tests from the virtual agent app's properties page in Virtual Agent Hub. You can: 

• Verify that the virtual agent app is correctly configured to communicate with the virtual agent provider.
• Add custom payload JSON to test the virtual agent's handling of custom payload data.
• Trigger a welcome intent event to test how the virtual agent begins a conversation.
• Trigger a custom intent event to test any custom events you've created for your virtual agent.
• Trigger a timeout event to test how the virtual agent responds when the contact times out.
• Test how the virtual agent handles DTMF Signaling tones that are generated when a user presses or taps a key on their telephone keypad. input.

The following image is an example of the properties page for a virtual agent app in Virtual Agent Hub. It shows the Configuration and Voice tabs and the Test pane.

1. In Studio, open or create a script and add a virtual agent action, if it doesn't contain one already.
2. Double-click the virtual agent action to open Virtual Agent Hub.
3. Click to select the configuration app for the virtual agent you want to test from the list on the left.

4. In the Test pane, enter a test message in the Type a message field and press Enter to send it. You can also click Microphone and speak into your computer's microphone.

5. Click Start with welcome intent to start a test conversation from the beginning. The virtual agent responds with its default welcome message. You can enter test messages.

6. Click More Options to test other aspects of your virtual agent.

   Learn more about fields in this step

   Field	Details
   Add Custom Payload JSON	Select to test sending custom payload JSON to your virtual agent. Enter the custom payload in the Add Custom Payload JSON text box that appears on the Test + Add page. Press Enter after each line. Send a test message when you're ready to test the custom payload. When you're finished testing custom payload, click Close on the Add Custom Payload JSON pane.
   Trigger Welcome Intent Event	Select to test the default Welcome intent The meaning or purpose behind what a contact says/types; what the contact wants to communicate or accomplish. The virtual agent sends the welcome message. Enter a message to respond.
   Trigger Custom Intent Event	Select to test a custom intent event. Enter the Event Name in the pop-up that appears and click Trigger. The virtual agent sends the message for the intent you entered. Send a response to test the intent.
   Trigger Timeout Event	Select to view the default timeout intent. The virtual agent sends the timeout response message that it's configured with.
   DTMF Input	Select to test your voice virtual agent's response to DTMF Signaling tones that are generated when a user presses or taps a key on their telephone keypad. tones. Enter DTMF Characters and click Input to see the virtual agent's response to those characters.

7. Click Reset to start your test again.

8. If you need to make changes based on the results of your testing, you can: 

   • Modify the Studio scripts for the virtual agent.
   • Modify the configuration settings in the virtual agent's app in Virtual Agent Hub.
   • Modify the virtual agent's configuration settings in the provider's console. Click the large square button on the app properties page in Virtual Agent Hub. The example at the beginning of this section shows the Google Dialogflow CX app with the provider button. It has the icon on it to indicate that the link opens an external site.
9. Test the virtual agent thoroughly after each change to your scripts or any configuration settings.
10. When you're finished testing, click Close.