The Talkative Configuration for your chatbot is very simple, and similar to configuring Dialogflow for any other service.
In this guide we will cover both Dialogflow Essentials and include any updated info or code for Dialogflow CX. An * indicates that a CX edit is following.
Dialogflow Configuration
- In Dialogflow, create a new ‘intent’ called “talkativeChatbotVerification”
- Make sure the text response is “verified”
Welcome Message
The first welcome message must have an event called “interactionStart”
Training Phrases
The user messages are sent verbatim to Dialogflow and they can be recognised via the Training Phases.
Responses
To respond to the message, simply fill in the Default Responses. One of those will be randomly selected as a response.
To fill in response suggestions, simply create a google assistant integration with suggestion chips, and fill in your desired responses*
*Dialogflow CX edit
To fill in response suggestions, send a custom payload as a response. The payload should look like:
Custom Actions
- Transferring To transfer, you will need to create a custom dialogflow response with the following content…
A transfer may fail for 3 reasons:
- businessHours
- queueCap
- noUsers
These can be set as events in DialogFlow to trigger follow on intents. You can also choose to ignore these within the DialogFlow custom payload, for example ➡️
To create such an intent, you will need to use these events ➡️
- Custom Messages based on Queue Availability You can offer different intents based on the state of queues.
This code snippet should go within the Custom Payload in your intent. Change the queue UUID to the Talkative queue you wish to reference.
queueStatusCheck_online
queueStatusCheck_offline_businessHours
queueStatusCheck_offline_noUsers
queueStatusCheck_offline_queueCap- Emails To allow the user to leave a message (that delivers as an email), you will need to send a custom response that contains ➡️ An email form will show to the customer in the widget. The data collection form will be the one set within the wait messages for that widget.
- Images To make the chatbot send a message that contains an image, you will need to send a custom response that contains ➡️
- Ending To end an interaction, you will need to send a custom response that contains ➡️
Ensure this is triggering as an ‘event’
To send multiple consecutive messages, simply create a second text response.
{
"talkative-engage": {
"action": "suggestions",
"suggestions": [
"suggestion 1",
"suggestion 2"
]
}
}{
"talkative-engage": {
"to": "<queue_uuid>",
"action": "transfer"
}
}You will need the UUID to match one of your queues.
{
"talkative-engage": {
"to": "<queue_uuid>",
"action": "transfer" ,
"ignore": [
"noUsers",
"businessHours"
]
}
}Exception Intents
If your transfer fails due to businessHours, queueCap, or noUsers, you can set up an exception-based intent.
interactionTransferFailed_businessHours
interactionTransferFailed_queueCap
interactionTransferFailed_noUsers{
"talkative-engage": {
"action": "check_queue_status",
"queue": "538afc06-d8ee-481e-9f4d-b2f08ba32cb4"
}
}You will then need to create intents that are triggered based on the four possible results of that queue presence check.
Each of these should be used as events to trigger the corresponding intent.
This allows you do have different welcome messages based on time of day for example.
Equally you could offer different hand-off journeys based on a queue presence check before getting to the transfer.
{
"talkative-engage": {
"action": "email"
}
}{
"talkative-engage": {
"action": "attach",
"url": "https://i.imgur.com/AYlbbmX.gif"
}
}{
"talkative-engage": {
"action": "end"
}
}Entities
Talkative automatically saves entities from Dialogflow into Interaction Data, allowing you to save key data points to your interaction record in Talkative. Both custom entities and standard system entities will be saved to Interaction Data. This happens automatically, no set up is needed.
Dialogflow Chatbot Intent Logs
You can view circumstances in which your Dialogflow chatbot has misunderstood a customer intent within Talkative.
- Via the Settings menu, under ‘Advanced’, select ‘Dialogflow Intent Logs’
Here you can examine interactions which have triggered a fallback intent message and a hand off to a customer advisor.
From here, you can take these customer messages and apply them as chatbot training data. To do this, you’ll need to create new intents with Dialogflow itself, incorporating the necessary data on the given topic.