Conversation Orchestrator can recommend bots that match customer intents in the Recommended Actions widget.

Follow along with this guide to set up the feature.

Create a bot

Conversation Orchestrator currently supports bots created in Conversation Builder, Google DialogFlow, and IBM Watson.

You will first need to enable the feature in Conversational Cloud. Please contact LivePerson support or your LivePerson customer representative to enable this feature.

Create Conversational Cloud Skill and User For Bot

Before adding automations to Conversation Orchestrator, you must first properly configure Conversational Cloud by creating a skill and a bot user.

The Conversation Builder Deploy the Bot tutorial guide explains how to create a skill and a bot user for each of your automations.

  • Create the skill names and bot user names that make sense for your use case
  • Be sure to match the bot skill to the bot user
  • Save the API credentials generated from your new bot user(s) as you will need them to connect your bot

Connect Bots to Conversational Cloud

Once Conversational Cloud is configured, your bot(s) need to be connected to enable them to respond to conversations in Conversational Cloud.

For Conversation Builder bots, see the next section.

For third party (Watson, DialogFlow) bots, see the section Configuring Third Party Bots.

Configuring Conversation Builder Bots

Follow the steps in the Conversation Builder Deploy the Bot tutorial to connect your bot.

Be sure to select allow Messaging conversations when you add the Agent for the automation.

Be sure to press the play button on the Bot in the bot dashboard to enable the bot to handle traffic.

For the next Configuring Conversation Orchestrator step, you will need the Bot ID. For Conversation Builder, this is called the Automation ID.

  1. Click the cog icon at the top of the Conversation Builder bot editor

  2. Select Automation Settings

  3. Scroll to the bottom and click on More Settings

  4. Copy the Automation ID and save this for the next section

Now skip to the next configuring maven section.

Configuring Third Party Bots

To connect a DialogFlow or Watson bot to LivePerson, you must follow the Bot Connectors Getting Started guide.

Then follow the specific guide for DialogFlow or IBM Watson

Be sure to press the play button in the bot dashboard to enable the bot.

For the next Configuring Conversation Orchestrator step, you will need the Bot ID. For DialogFlow, this is called the projectId. For Watson, this is called the Workspace ID.

Configure Conversation Orchestrator

Now that you have set up Conversational Cloud, and connected your bot(s), you need to link these bots to Conversation Orchestrator so they can be suggested in the Recommended Actions Widget.

Log in to Conversation Orchestrator

Log in to Conversation Orchestrator with your Conversational Cloud credentials and then navigate to Agent Assist > Recommended Actions - Automations.

Add the bots

Now use the interface to add the bots to Conversation Orchestrator. This enables Conversation Orchestrator to suggest these bots based on customers’ utterances, and add them to the conversation.

  1. Click the Add bot button to add a new bot

  2. Choose a Bot provider (Conversation Builder, Watson, or Dialogflow). Different bots support different types of credentials. For instance, DiaglogFlow uses a json object, whereas Watson provides the credentials in string format. This is reflected in the UI, when you change the bot type.

  3. Enter the bot name. Use the Login Name of the bot user created in Live Engage

  4. Enter the bot ID. The Bot ID is the value saved from the previous Connect Bots to Conversational Cloud section above and differs based on which bot platform you are using.

Next you will need to add the credentials from your bot platform into the Conversation Orchestrator Setup. If you are using a Conversation Builder bot, see the next section. For DialogFlow go here, and Watson go here.

Add credentials for Conversation Builder

Follow the steps below to gather information for the form.

  1. Login to Conversational Cloud and select the Automation Tab

  2. Click on the Account icon at the top right of the page, and choose Settings.

Click on the API tab, and copy the value listed for Your API Access Key. This should be entered in the Credentials field of the Add Bot page on Conversation Orchestrator.

Add credentials for DialogFlow

Follow the steps below to gather information for the form.

Let’s say that you already have a bot created in DialogFlow and we want to create a new credential.

  1. Click on the gear icon next to the bot name to get to the General Settings page.

  2. Click on the iam.gserviceaccount.com email address listed under ‘Service Account’.

  3. This will take you to the Google Service accounts page for your selected bot.

  4. Click on the email that was listed in Step 2. It should have a name of Dialogflow Integrations.

  5. Click Edit at the top of the page to start editing the service account details.

  6. Click Create Key at the bottom of this page.

  7. On the confirmation modal, select JSON key type, and click Create to save the credential file to your computer. Save this file, since you will not be able to download it again.

  8. Next, click on the IAM link in the left hand navigation.

  9. On this Permissions page, find the email listed in Step 2, and check that it has Dialogflow API Admin under the Role section. If not, edit the item and add the appropriate fields from the dropdowns.

Add credentials for Watson

Follow the steps below to gather information for the form.

Create a service, which generates a key which can be provided to LivePerson.

Next, create a skill and add intents.

The workspaceId can be obtained via the Skill Details, accessible from the Skills list.

The service can then request responses from Watson.

Configure Score Threshold

Automation suggestions have a score indicating how relevant the predicted intent is to a given consumer message. Suggestions are only shown if they are above the score threshold, which is configurable as shown below. The default score threshold is 70%.