1. Home
  2. Build with the Editor
  3. ActionBot Automation
  4. How To Create an ActionBot Conversation

How To Create an ActionBot Conversation

Updated on November 2, 2020 Download PDFDownload as PDF
Download PDF
Alternative Method Available

If you only plan to connect a Conversation to a single Smart Walk-Thru, consider building your Conversation from within the Smart Walk-Thru itself, as this method is quicker. Click here to learn how.

Brief Overview

This article explains how to build a Conversation which can be assigned to one or more ActionBots.

Conversations are built without any coding and allow you to answer the end-user’s question and/or collect information that later can be used to automate a process using a Smart Walk-Thru. Conversations can be used to complete common tasks, triage a support topic, navigate the end-user to a section of your application, or even provide a personalized search experience, all by using a simple chat interface.

To learn more about what the WalkMe ActionBot is, visit our article What Is the WalkMe ActionBot?

Steps For Building a Conversation

Step 1: Sign into your WalkMe Editor account

Step 2: Click the WalkMe apps dropdown triangle and select the ActionBots app

Step 3: Add a new Conversation

  1. In the Bots tab, select an already-existing Bot;
  2. Click the Assign Conversations field to open a drowndown menu;
  3. Click + Add new conversation:

Alternative method, from scratch

  1. Click on Conversations in the left sidebar;
  2. Click + ADD NEW CONVERSATION in the upper-right corner:

Step 4: Configure the question settings

  • On the New conversation page, fill in the following mandatory fields:

    • Name:
      • This is the private name (for identification purposes) that only you (and your fellow Conversation-builders) will see. The end-user will not see it.
    • Display Name:
      • This is the name your end-user will see if this Conversation is assigned to a Bot and appears in the chat dialogue box under the “Suggestions:” header text:
    • Keyword To Search By (Intent):
      • If the end-user types any of these keywords into the chat dialogue box of a Bot containing this Conversation, this Conversation will be triggered;
      • Enter as many keywords as you want. Simply type one, hit Enter, type another, hit Enter, etc.;
      • Pro-Tips:
        • Keywords are not case sensitive;
        • WalkMe recommends using analytics to see what end-users search for and updating your keywords accordingly.
  • Click SAVE.

Step 5: Create your Conversation’s flow

Create your Conversation flow using one or more of the Conversation building blocks. You must use at least one building block type, but it can be any one. None is mandatory per se.


    • Messages consist of text you write that a Bot may present to the end-user in the course of the Conversation;
    • Common uses of messages include providing end-users with useful links, explanations, and answers to questions.

    • Questions consist of text you write that a Bot may present to the end-user in the form of a question in the course of the Conversation;
    • The Bot will await the end user’s response to the question before continuing to the next step (which may be dependent on the end user’s answer);
    • You must select a type for each question. the options include Text, Boolean, Number, Regex, Phone, Date, Choice, and Choice – Salesforce:
      • Text Type Questions
        • No natural language processing (NLP) is used;
        • The ActionBot will log the end user’s entire input as the answer to the question and may re-use answers in follow-up questions.
      • Date Type Questions
        • This question type uses NLP. This means the ActionBot can interpret numerous date responses, e.g., “Tomorrow,” “Next week,” “1.2,” and “1-9-1990”;
        • You must select your preferred date format (DD/MM/YYYY or MM/DD/YYYY). This will determine the format used for the extracted answer;
        • The ActionBot can also extract date-time responses;
        • Advanced: You can configure the extracted date to a different format in a Smart Walk-Thru by using specialized values in the Smart Walk-Thru:

          Current Format Requested Format Value to Use in the Smart Walk-Thru
          MM/DD/YYYY (12/03/2018)



          MM/DD/YYYY (12/03/2018)



          MM/DD/YYYY (12/03/2018)



      • Boolean Type Questions
        • This question type uses NLP. The ActionBot will extract a positive or negative intent from the end-user and assign a true or false value as the extracted answer;
          • E.g., “Sure” would be extracted as true, and “Nope” would be extracted as false.
        • Boolean type questions are great for condition questions, as they’re easy to branch based on the value extracted.
      • Number Type Questions
        • This question type uses NLP. E.g., an end-user input of “five hundred dollars” will be extracted as 500.
      • Choice Type Questions
        • This question type doesn’t use NLP, but it uses a keyword search.
          • It will validate that only a value set in the Suggestions section can be selected by the end-user;
          • Additionally, you can assign keywords to each suggestion so that those keywords will select the suggestion as well.
      • Phone Type Questions
        • This question type uses NLP to extract any end-user input that could match a phone-number (using NLP models).
      • Regex Type Questions
        • This question type doesn’t use NLP.
          You can configure any question-type you could by using RegEx;
        • Insert the RegEx expression into the Pattern field;
        • You may also insert RegEx flags using any combination of the following:
          • g: Global match; find all matches rather than stopping after the first match;
          • i: Ignore case; if the u flag is also enabled, use Unicode case folding;
          • m: Multiline; treat beginning and end characters (^ and $) as working over multiple lines (i.e., match the beginning or end of each line, delimited by n or r, not only the very beginning or end of the whole input string);
          • u: Unicode; treat patterns as sequences of Unicode code points;
          • y: Sticky; matches only from the index indicated by the lastIndex property of this regular expression in the target string (and does not attempt to match from any later indexes).
        • Example:
          • Use Case: Build a question that validates that the end-user’s input is a valid email;
          • Solution: Using a Regex type question will allow you to validate that the input matches an email pattern;
          • Regex Value:  ^(([^<>()[].,;:s@”]+(.[^<>()[].,;:s@”]+)*)|(“.+”))@(([[0-9]{1,3}.[0-9]{1,3}.[0-9]{1,3}.[0-9]{1,3}])|(([a-zA-Z-0-9]+.)+[a-zA-Z]{2,}))$
      • Choice – Salesforce Type Questions

    • Conditions are used to branch the Conversation according to an end user’s answers to questions.

    • Actions enable you to trigger a WalkMe app or make a connection call during the course of the Conversation.
    • Selected WalkMe Items:
      • You can configure specific WalkMe items (of type Smart Walk-Thrus, Surveys, ShoutOuts, Shuttles, Resources, Live Chat, Open Menu or Connections) to be triggered during particular Conversations;
      • See examples using Smart Walk-Thrus and Connections below; the other Action types are the same, except for Open Menu, which just requires selecting the Action type.
    • Smart Walk-Thrus
      • The Smart Walk-Thru can be used to automate or guide the end-user through a process, redirect the end-user to a different location, or trigger an alternative resource for the end-user.

        The answers the end-user provides during the Conversation can be extracted and provided to the Smart Walk-Thru for automation purposes.

      • Connect a Conversation to a Smart Walk-Thru as follows:
        • Select a Smart Walk-Thru from the dropdown menu:
        • Click the + ADD FIELD button;
        • Select a question from the Data The Smart Walk-Thru Requires dropdown menu:
        • Create a variable name in the Name As field to the right of each question you select. Start each variable name with “${“ and end each variable name with “}”;
          • The final result should look like this: “${nameOfVariable},” with no spaces.
        • Use these variable names and format in the Smart Walk-Thru to which you want to connect this Conversation.
    • Connections
      • You can also configure specific connections to activate during the course of particular conversations. So besides the Smart Walk-Thru ‘attended’ automation you are now able to use ‘unattended’ automation via an API to save the user’s time and perform actions immediately in the background;

        The answers the end-user provides during the conversation can be extracted and provided to the Connection destination, for analytics and other purposes.

        1. Move to the Connections tab;
        2. Click ADD NEW DESTINATION:

        3. Fill details according to your desired connection destination:

        4. You have the ability to TEST your Connection destination before you create it;
        5. Click CREATE
          • Your new Connection destination should appear:
      • Connect a conversation to a Connection destination as follows:
        1. Select action type Call Connection from drop-down:
        2. Select your existing Connection destination;
        3. Important to make sure that all necessary properties have been added (the NameAs property should be automatically filled, make sure it is the same as your destination platform).
          • Data the WalkMe Connection Call Sends- The data you select will be sent to the Destination that was chosen under Action Type. If needed, rename the parameters to fit the destination system names.
          • Response Data Mapping- Define how to extract data from the JSON file received by the API and assign it to new variables. The variables will be saved and could be used later in the Conversation. It is important that the proper path is defined so that the correct data will be extracted and assigned to the variable.


    • You can use mentions (using the “@” symbol), or the value pulled from user responses, in both Questions and Messages. See the following example:
      • Add a new question, e.g., “Hello! What’s your name?”:
      • Add a new Message within the Message Text field and mention (using the “@” symbol) the previously created question at the point you want the end user’s answer to appear in the message:

        In this case, the value of the end user’s name will be pulled from his response to the question and inserted at this point in the Bot’s response.

        •  To mention a question, just type “@” and you will be presented with a dropdown list containing all of the available options:

Step 6: Connect your Conversation blocks together

  • Connect your Conversation blocks together by clicking the dropdown menu under Next step for each Conversation Flow item you have added and selecting what will occur after the current step plays.
    • If you’ve only added one item, select End conversation for that item.

Step 7: Publish your Conversation

  • Once you have added everything you want to your Conversation, click PUBLISH.
    • The PUBLISH button validates that the end user’s experience won’t be missing any crucial pieces (i.e., that all mandatory fields are filled, at least one Conversation flow item has been added, and all Next step lists items have been selected);
    • Once you click PUBLISH, a popup will appear in which you can see the current status of the Conversation, select the environment to which to publish the Conversation, and publish or un-publish the Conversation as desired:

  • Click PUBLISH in the popup window.
Best Practice

To create the best experience for your users, we recommend limiting your steps to the absolutely necessary, and no more than 100 steps in a conversation. More than 100 steps may affect the ActionBot performance.

Was this article helpful?

Related Articles

< Back