• 日本語

• 5.19 • 5.3 • 5.8

<translate>

How to Integrate Bots with Chat

Bright Pattern Contact Center integrates with providers like Alterra and IBM Watson to enable chatbots to be used in your contact center services. In this article, you will learn how to integrate an IBM Watson Assistant (i.e., a conversational bot) with your configured chat service.

In How to Configure Web Chat, you learned how to set up a chat scenario, chat service, and chat scenario entry point to work together to display a chat widget on your website. A working chat widget connects a customer on a website to a live agent in your contact center.

When a bot is integrated with your chat services, customers can be connected to the bot, which can answer questions, provide assisted self-service, and connect to a live agent if needed. Even if the chat is routed to an agent, the integrated bot still runs in the background, ready to assist the agent by providing suggestions (i.e., suggested responses for the agent to select and use).

Prerequisites

This article assumes that you have:

• Read and understand How to Configure Web Chat
• Already created a working chat widget
• Access to your company’s IBM Cloud account
• Some understanding of how Watson Assistant works

Procedure

This procedure consists of the following concepts:

• Create a bot in IBM Watson
• Add an IBM Watson integration account
• Create or choose a chat scenario that incorporates bot interactions
• Create or choose a chat service for your bot
• Link the chat scenario and service to your integration account
• Test the chat

Step 1: Create a Bot in IBM Cloud

1. If you haven’t already done so, sign up for an IBM account and launch IBM Cloud. IBM Cloud is the application where you will be developing and managing resources like Watson Assistant chatbots.
   
   
2. Follow all the steps in IBM’s Getting Started tutorial in order to create a Watson Assistant instance. The Watson Assistant will be your bot, and you will be adding intents, entities, and a dialog to it in a workspace. For more information on Watson Assistant, see IBM’s API Reference.

Step 2: Get Your Watson Credentials

Your credentials are necessary for integrating Bright Pattern Contact Center with your Watson Assistant. You will be using these credentials in later steps to add an integration account.

1. In your IBM Cloud console, get your Watson Assistant credentials from either of these two places:
   
   1. Watson Services > Assistant > Manage (for credentials in plain text)
      
   2. Watson Services > Assistant > Service Credentials (for JSON snippet)
      
      
2. Copy the Url, Username, and Password for your Watson Assistant to a separate doc, such as a text file.

Watson Assistant credentials

Step 3: Add an Integration Account

Integration accounts are what allow your contact center to operate with third-party services.

1. In the Contact Center Administrator application, go to Call Center Configuration > Integrations Accounts.
   
   
2. Click the Add account (+) button to add a new integration account.
   
   
   

   Add new integration account
   
   
   
3. Select account type Bot / Chat suggestions engine and click OK.
   
   
   

   Select "Bot / Chat suggestions engine"
   
   
   
4. In the Bot / Chat suggestions engine type dialog, select Watson Conversation and click OK.
   
   
   
   
   
   

Properties

Integration account properties

Fill in the following fields:

• Name - A unique name for your bot
  
  
• Type - By default, “Watson Conversation” because you selected this type when adding the account
  
  
• Url - The endpoint of your Watson Assistant (https://gateway.watsonplatform.net/assistant/api) + /v1/workspaces/ + your workspace ID (see below) + /message?version=2017-05-26
  
  • For example: https://gateway.watsonplatform.net/assistant/api/v1/workspaces/085c97f9-44c3-403d-8aad-92e86d9379f8d/message?version=2017-05-26
    
  • Note: You must enter the Url in this way for your bot to work properly.
    
    
• Workspace ID - The identifying number of your Watson Assistant Workspace
  
  • To find this ID, go to IBM Watson Assistant > Workspaces, click on the desired workspace’s Actions button, and select View details.
    

IBM Watson Assistant > Workspaces

• • Copy your workspace ID from there.
    

Copy your workspace ID

• Username - The username for your IBM account that you copied from your service credentials
  
  
• Password - The password for your IBM account that you copied from your service credentials
  
  
• Max API calls per day - The maximum number of calls that can be done each day; this limit is here to keep your account from being charged for additional calls beyond the first 1,000 that are included in a free account.
  
  
• Maximum suggestions - The maximum number of suggestions (e.g., 3) that can be delivered to the Agent Desktop during active chat interactions. Suggestions are the bot-generated replies that agents can select and use during chats.
  
  
• Test Connection - Click to be sure Bright Pattern Contact Center can connect to your Watson Assistant.
  
  • If the credentials are invalid, go back to the Url property and make sure you’ve entered it exactly as explained in this procedure.
    
  • Test until you see the success dialog: "Account credentials appear to be valid."

Click Apply at the bottom of the screen to save your account properties.

Step 4: Create or Select a Chat Scenario

Note that this step is the same as Web Chat Configuration Step 1.

1. Go to Configuration > Scenarios > Chat.
   
   
   

   Configuration > Scenarios > Chat
   
   
   
2. Either select an existing Chat Scenario from the list, or click the Add from template button at the bottom of the screen to create a new chat scenario from the “Mobile Chat” template.
   
   
   

   Select the "Mobile Chat" template
   
   
   
3. Creating a new chat scenario from a template will open the Scenario Builder application in a new browser tab or window. For the purpose of this simple setup, leave the scenario as-is and click Save.
   
   
   

   Name the scenario
   
   
   
4. Give the scenario a unique name (e.g., "Bot Scenario") and click Save. Your new scenario will appear in the list of scenarios.

Step 5: Edit the Scenario to Work with Your Bot

To use the Watson Assistant in live web chat interactions, you’ll have to edit the basic Mobile Chat scenario template to allow the bot to interact with the customer. For more information about scenarios, see the Scenario Builder Reference Guide.

There are many ways to construct such a bot-enabled scenario; the instructions given in this step are provided as just one example of how to use your Watson Assistant in chat.

In this scenario-building exercise, you will be configuring the system to use the bot to:

• Automatically accept the customer’s chat request
• Greet the customer
• Provide answers to the customer
• Provide suggested replies to the agent
• Connect to an agent when it’s determined that the bot is not being helpful

Example chat scenario in Scenario Builder

1. Download File:App Example Bot Scenario.zip and import it to your Contact Center Administrator application.
   
   
2. Select the imported scenario and click EDIT to open it.

The following scenario blocks (with comments) are included in this example scenario. Be sure to read the comments that explain each scenario block's properties.

1. Chat Bot Select Account: This block tells the system which integrated bot will be used for this scenario.
   
   
   

   Select an integration account
   
   
   
2. Set Variable: This establishes the customer's first name and last name. Setting this context helps to personalize the conversation.
   
   
   

   Use a variable to get the customer's name from the pre-chat form and use it in the chat interaction
   
   
   
3. Send Message+: This block delivers a chat message to the customer, incorporating the customer's name.
   
   
   

   Have the bot send a message to the customer, addressing him/her by name
   
   
   
4. Set Variable: This time, we are defining a step. In this example scenario, we want the bot to respond to the customer two times before connecting to the agent. The first time is specified as "step 0" and the second time is specified as "step 1." The scenario will run through both steps. If the customer's issue is not resolved by step 1, the customer is transferred to the agent for help.
   
   
   

   Use a variable to define this part as the first in a series of steps
   
   
   
5. If: The If block specifies what happens if the customer did not write a message in the pre-chat form's Message field. If that field is empty, the scenario tells the bot to send a message. The message aligns with what you have configured your Watson Assistant to say. We call the bot's messages suggestions.
   
   
   1. Add a branch
      
      
   2. Give it a condition of item.message is empty
      
      
      

      If the customer's message is empty, have the bot send a message to the customer
      
      
      
6. Ask a Bot: The scenario uses this block to get suggestions (i.e., some response from the Watson Assistant) to deliver to the customer. For each of the conditional exits (Failed, Timeout, or No Data), the scenario will send a message and then connect to the agent.
   
   
   

   Get suggestions from the bot
   
   
   
7. Send Message+: This is used to send a message to the customer via either chat or SMS. In this example, the message is sent in the active chat interaction. If a specific message is not defined here, the standard system-wide message is sent to the customer.
   
   
   

   Send another message if you want
   
   
   
8. Set Variable: This time, we are defining "step 1." Remember, we want the bot to respond to the customer two times before connecting to the agent. The first time is specified as "step 0" and the second time is specified as "step 1." The scenario will run through both steps. If the customer's issue is not resolved by step 1, the customer is transferred to the agent for help.
   
   
   

   Define this part as step 1
   
   
   
9. If: In this If block, a condition is set to make sure that if the scenario takes more than two steps to resolve the customer's issue, then it will transfer the chat from the bot to the agent.
   
   
   1. Add a branch
      
      
   2. Give it a condition of “steps” > “2”
      
      
      

      If the scenario takes more than two steps to resolve the customer's issue, then the agent steps in
      
      
      
10. Find Agent: The Find Agent block looks for the next available skilled agent to accept the chat. You can use Find Agent to set wait times and send the customer messages about estimated waiting time (EWT).
    
    
    

    The bot couldn't resolve the customer in two steps, so the scenario looks for an agent
    
    
    
11. Connect Chat: Connect Chat is for setting the destination of the chat. If you leave the destination fields empty, the Find Agent block will find the first available agent.
    
    
    

    You can either specify a destination for the chat or let the system find the next available agent
    
    
    
12. Exit: The Exit block completes the scenario. Without it, the scenario will loop through this configured flow until the customer ends the chat.

Step 6: Create or Select a Chat Service

See Web Chat Configuration Step 2.

Step 7: Create or Choose a Chat Scenario Entry

Note that this step is basically the same as Web Chat Configuration Step 3, with the addition of another important property.

1. Go to Configuration > Scenario Entries > Messaging/Chat and select a scenario entry point to use with your chat scenario. The entry point is what starts the scenario.
   
   
   

   Messaging/Chat scenario entries list
   
   
   
2. If there are no existing chat scenario entries shown, you have to create one.

How to Add a New Chat Scenario Entry

1. Click the Add chat scenario entry button.
   
   
2. In the Properties tab, fill in the properties as appropriate (see below for required and optional properties). For definitions, see the Contact Center Administrator Guide, section Messaging/Chat.

Required Properties

In the Properties tab, make sure to address the following fields right now. These are critical to chat configuration.

Messaging/Chat scenario entry properties that you need to address for chat to work

• Name - The unique name (any) of the chat scenario entry (e.g., "Redwood Chat")

• Unique identifier - A long string of characters that identifies the chat entry. It is recommended that you change it to something short and unique (e.g., "1234").

• Scenario - Choose the Chat Scenario you selected or created earlier in this procedure (e.g., "Super Chat"). This is where the parts come together to make chat possible.

• Service - Choose the Chat Service you selected or created earlier in this procedure (e.g., "Pacific Chat"). This is where the parts come together to make chat possible.

• Bot / Chat suggestions engine - The engine that will provide suggested replies to the agent during chat. Note that in order to use a bot/chat suggestions engine, you must have an integration account configured for your contact center.

Other Properties

The other remaining properties are optional at this time. You can come back to them later.

Step 8: Edit Chat Widget Style

See Web Chat Configuration Step 4

Step 9: Add a Contact Tab

In order for this chat scenario entry to work, you must add a contact tab for the chat widget. If you don’t do this, your webpage will not display the chat widget!

1. Click on the Chat widget tab.
   
   
2. Beneath Chat Initiation via Contact Tabs, click add.
   
   
   

   Add a new contact tab
   
   
   
3. The Chat Widget Configuration editor will open in a new browser tab or window. Edit the properties and fields as desired.
   
   
   

   Chat Widget Configuration editor in new browser tab/window
   
   
   
4. Required: At the most basic level, these actions are required right now:
   
   * Name - The unique name (any) for this new contact tab
   
   * Pre-Chat Form Configuration > Fields - Common: Back in Step 5, you specified variables to get the customer's first name and last name and use them in chat messages. In order for them to work, you must enter fields first_name and last_name here. For example:
   
   
   

   Click "add" to add a new common field
   
   
   * Save - Click to save changes
   
   
5. Back in the Chat widget tab, go to Chat Initiation via Contact Tabs.
   
   
6. Beneath the contact tab you just added, click add url.
   
   
   

   Make sure you add the URL of your webpage
   
   
   
7. Enter the URL of the website where the chat widget will be placed, and click the green checkmark to apply your changes.
   
   
   

   Apply changes or cancel
   
   
   
8. Optional: At this time, you may add or edit any of the other Chat widget tab sections. You can always come back to this for further changes.
   
   
9. At the bottom of the screen, click Apply to save your changes.

Step 10: Get the HTML

See Web Chat Configuration Step 6

Final Step: Test the Chat

You have now successfully set up your contact center service, scenario, and chat widget to work in an integrated manner with your Watson Assistant. You should test it to make sure that the:

• Customer's name entered on the pre-chat form comes through during the chat
• Bot is able to automate the conversation with the customer
• Customer is connected to an agent if the bot couldn't help within two tries
• Agent is able to see the bot's suggestions during the active chat

The following screenshot shows a service chat (on the Agent Desktop side) where the bot provides two responses to the customer and then connects to an agent. You can see the bot's suggestion below the agent's text input field.

Example chat as seen on Agent Desktop

And the following is how the chat looks to the customer:

Example chat as seen by customer on webpage

</translate>

< Previous | Next >