How to Setup Conversational AI

Conversational AI is a user-friendly menu system in which a customer can interact through keywords and will dynamically route their calls to its proper destination.

While using this feature, the customer will have a fast, automated conversation with the IVR, ensuring that only qualified customers will get through to your agents.

Upload the Audio Files needed for Conversational AI.

See this link to know more about how to do it.
How to Upload Audio Media

Creating a Conversational AI Script

From the Dashboard, click on Apps and select Conversational AI from the drop down.

Click on the Add Conversational AI button.

Control Panel Section will open.

Type the Script Name of this Conversational AI Script

Right-click on the Blue Element Box Question and click on the Edit Option.

Edit Box Section will appear. Configure the following.

Start Point - This will be the start point of the Conversational AI.
Answering Machine Detection - If this option is On, the system will filter Answering Machine Calls. (Turned Off by default)
Name: Type a descriptive name for this Question.
Audio Sensitivity: This section is the ability of the system to detect the customer's voice.
- If the customer's audio has a lot of background noise, Audio Sensitivity should be lower to filter out background noise.
Speech Timeout - This where we set how long the system will listen for a reply before executing an action.
- Example scenario: If we set the speech timeout to 500ms (5sec) and the person talked for 10 seconds, the system will only listen for the first 5 seconds of what the person said.
Audio File: This section is where we select the Audio Message that customer will hear. The audio file usually contains instructions on what the customer will say or do.
Skip Speech Detection - If this feature is ON, the system will play the audio recording set on the question and will route straight to the destination after the audio recording played.
Speech Input Delay: Determines how soon a customer can respond to the question.
- For example, we set the delay to 5 seconds. When the customer hears the audio message and responds before 5 seconds, the system won't be able to detect his/her response. He/She needs to wait 5 seconds before they can react.
- If we set the delay to "Delay Until Audio is Played," the customer won't be able to respond until he finishes the entire message.

Invalid Response Audio Sensitivity: This section is the ability of the system to detect if the customer responded with an invalid response.
- If the customer's audio has a lot of background noise, Audio Sensitivity should be lower to filter out background noise.
Invalid Response Speech Timeout - This where we set how long the system will listen to determine if the response is valid or not. If the answer is invalid, it will trigger the Invalid actions.
- Example scenario: If we set the speech timeout to 500ms (5sec) and customer talked for 10 seconds, the system will only listen for the first 5 seconds of what the customer said and determined if its a valid response or not.
Max Invalid Response - This is where we the set limit of Invalid Responses before executing the Invalid Response Actions.
Invalid Responses - Select the audio file that will play if the customer responded with an invalid response.
Invalid Destination - Select a destination where you want to send the call if the customer responded with an invalid response. (Click View button to see the Destination)

Example Destination Options:

Campaigns - Select a campaign where the call will be transferred to any available agent on that campaign.

No Response Audio Sensitivity: This section is the ability of the system to detect if the customer responds or not.
- If the customers audio has a lot of background noise, Audio Sensitivity should be lower to filter out background noise.
No Response Speech Timeout - This where we set how long the system will listen to determine if there is a response or not.
- Example scenario: If we set speech timeout to 500ms (5sec), the system will decide after 5 seconds if the call is considered No Response.
No Response Timeout - After validating that the call is a No Response, this section is where we set how long the system will wait before executing the no response actions.
- Example scenario: If we set No Response Timeout to 10 seconds, and the call is already considered a No Response, the system will trigger the No Response actions after 10 seconds.
Max No Response - This is where we the set limit of no response instances before routing the call to the No Response Destination Section
No Response Audios - Select the audio file that will play if the customer did not respond.
No Response Destination - Select a destination where you want to send the call if no response detected. (Click View button to see the Destination)

Click on Apply Button to Save the Changes.

Note: System will not allow you to Save or Apply Changes if Audio Settings is not configured.

Cloning a Question Element

To Clone or Copy Question Elements: Right-click on the Question Element and click on Clone.

The system will create a copy of the cloned Question Element and will be indicated on the element's name.

Adding A New Question Element

To add a New Question Element, click on the Add Question Button.

A new Question Element will be added to the script.

Destination Element Configuration

Next up is to configure the Destination. This is where the call will go from the Connection Source to Recognition to Destination.

Right click on the Destination box and select edit.

On the Edit window, configure the following:

Name - Name of the Destination.
Recording - Select a Recording that you want to play once it goes to the Destination.
- Click on Upload File to add a new Audio File from the computer.
Route To - Select where you want to route the call. (Click View button to see the route to settings)
Disposition - Select the status that the system will use after the call was routed.
DTMF - Select the DTMF Digit that the system should select once routed to this Destination.

Click on the Apply button to save the settings.

Note: If the Charge is set to No, Conversational AI settings cannot be saved unless the Route To Action is set to Terminate Call

This message window will appear

Adding A New Destination Element

To add a New Destination Element, click on the Add Destination Button.

A new Destination Element will be added to the script.

Recognition Element Configuration

A Recognition Element is the connector of Elements. It can connect a Questions Element to another Question Element and a Question Element to a Destination Element.

To edit a Recognition Element, right click on it and click on Edit

Under the Edit Connection window, you can now put the name of the recognition and the recognize words/values.

Add the word or value that will trigger the system to transfer you over to the destination.

Name - Name of the Connection.
Upload Options - To upload recognize words or values
(Click here to know more on How to Upload, Copy and Export Words or Values.)
Copy Options - Copy the same recognize words/value from the other connection.
Store Match to Lead Field - Recognize Word/Value will store on the selected field.
Add Recognize Words/Values - Enter the words or values you want to use.
AI Training - search the most common responses to questions for the purpose of training the AI. It will compile the top 50 responses and display how many times that response was recorded.
- Select any Date range and click on the Train button.
- Click on the drop-down to view the most common responses that were recorded by the system.
- You can add those common responses as your connector by selecting the word on the drop-down option then select a type
- Click on Add button.

To Configure the Responses

Select the Type of Recognize Words or Values with either Contains, Exact Match or Regular Expression. You can also delete an entry by clicking on the Remove button.

You can also set the priority of each word/values based on the priority level. The higher the number, the higher the priority.

Down below you will see the connection source and where it will go once the system accepts your words/values recognition.

Click on the Apply button to Save the changes.

Note: System will not allow you to connect a Fixed recognition to an existing destination.

If you want to have a new recognition to connect it to the same destination, you need to drag a new recognition from the Question and drop it to the destination.