This documentation supports the 25.3 version of BMC Helix Digital Workplace Basic and BMC Helix Digital Workplace Advanced. Icons distinguish capabilities available only for the Advanced and External license levels. For more information, see License types and features. To view an earlier version, select the version from the Product version menu.

 

Using BMC HelixGPT to make services requestable via chat

Catalog administrators can make services requestable via chat so that end users can request such services in a conversational manner via Employee Navigator. Catalog administrators can use one of the following methods to make services available via chat: 

Using BMC HelixGPT to make services requestable via chat provides the following benefits to the catalog administrators: 

  • Save manual efforts required for using the Publish chat-enabled service wizard.
  • Avoid human errors when using disambiguation, providing variations in service descriptions, service clarification phrases, and natural language variations of the questions.
  • Achieve faster scalability by making several services requestable in a short period of time. 
  • Improve discoverability of services that help users identify the appropriate service for their queries. 
Important

Services that include a questionnaire with the following types of questions are not supported:

  • Table (multi-select)
  • Table (single-select) 

Catalog Request Agent

The Catalog Request Agent, which is configured in HelixGPT Agent Studio includes the prompts to make services requestable. It identifies and suggests the services relevant to the user queries. After the end user approves, the agent submits the service request on behalf of the user. To learn how to edit an agent, see Managing AI agents.

The Catalog Request Agent is provided out-of-the-box with BMC HelixGPT, as shown in the following image:

Catalog Request Agent

Creating a copy of the Catalog Request Agent and modifying the prompts

If you want to modify the Catalog Request Agent setting, make sure you create a copy of the agent. For more information, see Managing AI agents.

Important

To make sure that the services are suggested, after creating the agent, make sure that the prompt has the following structure. All prompt sections must be present and the individual prompts cannot be blank.

 

Prompt structure

    ## Search Service

    <Prompt for creating a query to search the services based on user's query>

    ## Best Service

    <Prompt for selecting the most relevant service from the list of services>

    ## Predict Answers

    <Prompt for predicting answers to the questionnaire based on user's query>

    ## Question

    <Prompt for selecting question to ask>

    ## Failed Question

    <Prompt for asking the question again after the user failed to answer it>

    ## Format Answer

    <Prompt for formatting the user's answer (like splitting to multiple answers or skipping the question completely)>

    ## Answers Confirmation

    <Prompt for analyzing if the user wants to submit request or change any answer>
 

 

(Optional) To configure Catalog Request Agent

As an administrator, you can modify the number of services, questions, and probable answers that the Catalog Request Agent provides to the end users. 

  1. Log in to BMC Helix Innovation Studio.
  2. From the Workspace tab, select HelixGPT Agent Studio.
  3. On the Records tab, select the checkbox next to Agent and click Edit Data. 
  4. Select the record instance for the copy of Catalog Request Agent that you have created, and click Edit. 
    In the Configuration field, specify the following parameters:
    ParameterDescription
    services_limit

    Controls the maximum number of services that the system searches in the catalog. Among the searched services, the most relevant service that match their query is displayed to the end users .

    Default value: 10

    questions_limit

    Controls the maximum number of questions that a service can have. If a service has more questions than the number specified in this parameter, then the user is prompted to use the service request submission UI. 

    Default value: 20

    predict_answers

    Determines whether initial user input queries are used to predict answers to the questions.
    For example, if a user types I need to request guest parking from Monday to Wednesday next week", the system predicts the start and end dates for the parking request. 

    If you do not want to predict the answers, enter false.

    Default value - true

    restrict_servicesRestricts the service version IDs from being submitted by the end users. 

    The following image is an example of the parameters that you can configure for the Catalog Request Agent:
    Catalog request agent config in IS.png
     

  5. Click Save. 

Result

After the Catalog administrator has completed the configurations, the end user can submit the service requests via chat, as shown in the following image: 

Submit requests via chat

 

 

Tip: For faster searching, add an asterisk to the end of your partial query. Example: cert*