Enlighten Autopilot Knowledge

Autopilot Knowledge combines digital conversational botsClosed A software application that handles customer interactions in place of a live human agent. from CXone Bot Builder with the knowledge resources from CXone Expert. CXone Expert enables your bot to answer customer questions instantly. You set up your knowledge baseClosed A website that stores troubleshooting articles. in CXone Expert and connect it with Bot Builder. This helps you save time on implementation. After you configure your bot, deploy it to the desired digital channels.

Prerequisites

To use Autopilot Knowledge, you must have:

Set Up Autopilot Knowledge

Complete each of these tasks in the order given.

Add Autopilot Knowledge from the Skill Store

The skill store is a public repository of approved bot skills.

  1. In CXone, click the app selector and select Bot Builder.

  2. Click Go to skill store and select Autopilot Knowledge.

  3. Click Add to your bot and select your bot from the drop-down.

Connect Your CXone Expert Knowledge Base

  1. Go to your CXone Expert portal and gather the following information:

  2. Navigate to your Autopilot Knowledge integration in Bot Builder:

    1. In CXone, click the app selector and select Bot Builder.

    2. Select the bot you configured for Autopilot Knowledge.

    3. Click Integrations icon, which looks like a plug. in the left icon menu.

    4. Click the Scripts tab.

    5. Select Autopilot Knowledge.

  3. Under Variables, click Expert_URL and enter the URL you copied from your CXone Expert portal.

  4. Click Expert_API_Key and enter the API key you generated.

  5. Click Expert_API_Secret and enter the secret you generated when creating the API token. Changes are saved automatically.

Fine-Tune the Autopilot Knowledge Integration

Follow these steps to configure optional variables to fine-tune Autopilot Knowledge.

Variable Details
Persona

Holds a text-based persona description. This description provides Autopilot Knowledge with its role and purpose. For example, "You are a friendly bot that answers questions about billing." This improves bot performance and customer experience by strengthening brand identity and humanizing automated interactions.

Expert_Kernels

Holds a number, 1 through 5. This is the number of kernelsClosed Pieces of CXone Expert articles that are relevant to a contact's issue. you want Autopilot Knowledge to use when generating responses. This enables Autopilot Knowledge to target the parts of your CXone Expert knowledge base that are relevant to the contact's message. The default is 3.
Expert_Threshold Holds the decimal form of a percentage, 0.5 through 0.9. This is the required similarity index between the contact's message and the CXone Expert kernel. For example, if you enter 0.85, Autopilot Knowledge doesn't use kernels that are less than 85% a match for the contact's issue. The default is 0.6.

  1. If you haven't already, navigate to your Autopilot Knowledge integration in Bot Builder:

    1. In CXone, click the app selector and select Bot Builder.

    2. Select the bot you configured for Autopilot Knowledge.

    3. Click Integrations icon, which looks like a plug. in the left icon menu.

    4. Click the Scripts tab.

    5. Select Autopilot Knowledge.

  2. Click Persona and enter a short, specific persona description that tells Autopilot Knowledge its role and purpose. For example, "You are a friendly bot that answers questions about billing." This helps Autopilot Knowledge provide more relevant responses.

    Make sure your persona description is concise, specific, and easy to understand. A poorly written persona description may result in poor responses from Autopilot Knowledge.

    • Good: "You're a technical expert answering questions about our appliance company, OneAppliance. OneAppliance sells and services washers, dryers, and dishwashers. Be succinct and use technical language." (Provides context and specific instructions.)

    • Poor: "You're an expert on OneAppliance. Answer all customer questions." (Doesn't provide context. Instructions are very general.)

    • Good: "You're a banking expert answering questions about our bank, OneBank. Common topics include personal finance, investing, and taxes. Your specialty is taxes. Be as concise as possible." (Specific but simple.)

    • Poor: "You're an extremely knowledgeable bot working for OneBank, the world's leading financial institution for personal asset management, such as savings accounts and certificates of deposit, and private equity, funds for acquiring private businesses. Your objective is to provide information in a succinct manner, enabling customers to accomplish their financial goals." (Long, complicated, and difficult to understand.)

  3. Configure variables to customize how Autopilot Knowledge uses CXone Expert kernels . Kernels are pieces of articles that are relevant to a contact's issue.

    1. Click Expert_Kernels and enter the number of kernels you want Autopilot Knowledge to use when generating responses. For example, if you enter 2, Autopilot Knowledge only uses the top two kernels that relate to the contact's issue. Enter a number between 1 and 5.

    2. Click Expert_Threshold and enter the decimal form of the match percentage you want to require for kernels. For example, if you enter 0.85, Autopilot Knowledge doesn't use kernels that are less than 85% a match for the contact's issue. Enter a number between 0.5 and 0.9.

Configure Variables for Error Messages

To customize the messages Autopilot Knowledge sends to contacts when it experiences an error, configure the variables for error messages.

  1. If you haven't already, navigate to your Autopilot Knowledge integration in Bot Builder:

    1. In CXone, click the app selector and select Bot Builder.

    2. Select the bot you configured for Autopilot Knowledge.

    3. Click Integrations icon, which looks like a plug. in the left icon menu.

    4. Click the Scripts tab.

    5. Select Autopilot Knowledge.

  2. In each error message variable, enter the text you want Autopilot Knowledge to display for the error message. The variables that appear and their names depend on your version of Autopilot Knowledge.

    Depending on your version of Autopilot Knowledge, variables may appear for these errors:

    • The contact's message is too long.

    • The contact's message is empty.

    • No available knowledge. This means that there is no content related to the contact's issue in your CXone Expert knowledge base.

    • No relevant knowledge. This means that Autopilot Knowledge was not able to generate a final response though it found kernelsClosed Pieces of CXone Expert articles that are relevant to a contact's issue. in CXone Expert related to the contact's issue.

    • Couldn't connect to CXone Expert.

    • Invalid CXone Expert credentials.

    • The utterance is too short. This means that the contact's message did not meet the minimum length requirement (more than two words).

Configure Debug Variable

To see the back end of Autopilot Knowledge as it processes customer issues, set the Debug variable to True. This helps you identify gaps between the questions contacts are asking and the answers available in your CXone Expert knowledge base.

With Debug set to True, Autopilot Knowledge displays the results it gathers from CXone Expert and the raw output it generates. This includes the kernelsClosed Pieces of CXone Expert articles that are relevant to a contact's issue. it found. Both display in the console of the Generate Reply and Suggest Articles bot action in Bot Builder.

  1. If you haven't already, navigate to your Autopilot Knowledge integration in Bot Builder:

    1. In CXone, click the app selector and select Bot Builder.

    2. Select the bot you configured for Autopilot Knowledge.

    3. Click Integrations icon, which looks like a plug. in the left icon menu.

    4. Click the Scripts tab.

    5. Select Autopilot Knowledge.

  2. Set the Debug variable to True or False.

Configure the Generate Answer and Suggest Articles Bot Action

The Generate Answer and Suggest Articles bot action determines:

  • The bot's target word count for replies.

  • If and how the bot displays suggested CXone Expert articles to the contact.

  1. In CXone, click the app selector and select Bot Builder.

  2. Select the bot you configured for Autopilot Knowledge.

  3. Click Dialogues icon, which looks like conversation bubbles. in the left icon menu.

  4. Click the Fallback tab.

  5. Under Fallback, click Action.

  6. Adjust any of the parameters of the Generate Reply and Suggest Articles bot action.

    Field Details
    Generated Reply - Word Count

    How long every generated reply should be. Autopilot Knowledge tries to fit every reply to the desired word count. Consider that the larger the word count, the longer it takes for the contact to read and respond. So, select a word count that correlates to your desired response time.

    The default value is 100. If you set the value to 0, reply generation is disabled. There is no minimum or maximum value.

    Suggested Articles - Format

    How you want suggested articles to display. Options include Buttons, Cards, and None. If you select None, the bot cannot send suggested articles. You may want to select this option if you only want the bot to reply to questions. You can further customize the formatting if you have basic knowledge of JavaScript.
    Suggested Articles - Max Count The maximum number of suggested articles the bot can send to the contact. The default value is 5. There is no minimum or maximum value.
    Suggested Articles - Message Content The text you want to display when the bot sends suggested articles to the contact. For example, you might enter Here are some articles related to your question that you may find useful.

Configure Additional Bot Actions, Integrations, and Other Features

You can customize your bot through the features in Bot Builder. You can customize bot actions with script integration, but this method requires basic knowledge of JavaScript. You can also disable or edit variables to display in bot actions.

Test Bot Connection

CXone Bot Builder lets you chat with your already-created botClosed A software application that handles customer interactions in place of a live human agent. to help it learn. Talk to your bot to ensure it works the way you want it to.

  1. In CXone, click the app selector and select Bot Builder.

  2. Click the bot you want to work with.
  3. From any page, click Talk to your bot icon, represented by a conversation bubble.
    The Talk to Your Bot screen.
  4. Enter a messageClosed Anything a contact says in a bot interaction, whether question or statement, written or spoken. in the field at the bottom of the chat window, just as you would in any chat conversation. Your bot responds as it would in a real interaction.
  5. Continue to enter messages and review the bot's responses. If at any point the bot responds in an unexpected way, you can stop and correct the issue. For example, you might need to create a new intentClosed The meaning or purpose behind what a contact says/types; what the contact wants to communicate or accomplish or ruleClosed Used to define bot's response to messages that don't change with context..
  6. If you need to restart the interaction after correcting an issue, click Start New Conversation.

Deploy Bot to Digital Channels

When you're finished configuring your bot and you're ready to put it into production, follow these steps:

  1. Verify that the CXone employee profile you created for your bot is assigned to the digital skill or routing queue that will route contacts to your bot.
  2. Set up a safety net, if you haven't already done so.
  3. Deploy the bot model to production.

After your bot is deployed to production and online, the deployed bot modelClosed Version of a bot that has been trained and staged will behave just as live agents do when they are online. CXone assigns new incoming cases to the bot's digital inboxClosed Area where cases appear in the digital interaction workspace in an agent application from the digital skills or routing queuesClosed Digital First equivalent of a skill; routes each contact to an agent assigned to handle that type of contact the bot is assigned to. If your bot is deployed to production and is online, but is still not receiving cases, you may need to check the routing configuration.