Generating AI-retrieved answers for end users through BMC HelixGPT
Benefits of using BMC HelixGPT
End users can benefit from generative AI retrieval of answers in the following ways:
Answers are extracted from specific knowledge articles, | Save time in going through multiple service requests or approvals to get targeted answers in a conversational format. |
| The conversation is retained across screens in the end-user console. |
Difference between search, chatbot, and generative AI retrieval of answers by usingBMC HelixGPT
The following table describes the difference between search, chatbot, and BMC HelixGPT:
Search | BMC Helix Virtual Agent | BMC HelixGPT |
---|---|---|
When a user enters search keywords, the results (default up to 25) that are the closest match to the search text are displayed, irrespective of their source or type. | When an end user enters a question, BMC Helix Virtual Agent understands the intent of the user and provides an answer based on the configured knowledge sources. The chatbot also asks questions required to submit a service request on behalf of the end user. | When a user enters a question, the system understands the question and generates a summary of the relevant answers from all the available knowledge sources, which is presented as a human-like text response. (Version 23.3.04 and later) If configured, BMC HelixGPT also offers to submit a default service request on behalf of the end user. |
Search engines index and retrieve data by using keyword-based indexing. | The Catalog services are imported into BMC Helix Virtual Agent, which are then provided as suggestions to the end users. | Generative Pre-trained Transformer (GPT) uses large language models (LLMs) to understand the intent, context, and semantics of human language. |
In BMC Helix Digital Workplace, end users can use full-text search from the following areas of the end-user console:
| In BMC Helix Digital Workplace, end users can click the chatbot icon available on all screens to start interaction. Important: You can enable only one self-help application - chatbot or BMC HelixGPT for the end users. | In BMC Helix Digital Workplace, end users access the Employee Navigator to ask questions from the following areas of the end-user console:
Important: You can enable only one self-help application - BMC Helix Virtual Agent or BMC HelixGPT for the end users. |
Workflow to configure BMC HelixGPT with BMC Helix Digital Workplace
The following image provides an overview of the workflow to configure BMC HelixGPT with BMC Helix Digital Workplace:
The following table provides the detailed description and references to complete the workflow:
Task | Action | Description | Reference |
---|---|---|---|
1 | Request this capability | Contact your BMC account manager to get this capability. | Not applicable |
2 | Complete the prerequisites | Complete the prerequisites such as deploying a Large Language Model (LLM) service from a supported vendor of your choice, get the API key and other config parameters. | |
3 | (If you want to leverage the out-of-the-box chat bar in the My Activity or Catalog pages) Set the default Skill | Specify the default Skill that you want to use to include BMC HelixGPT chat bar in the My Activity and Catalog pages. A Skill is a container for the artifacts that define the knowledge sources to be referenced. The Skill is associated with the GPT provider. A default prompt for generating answers from knowledge sources is also provided with the default Skill. | |
4 | (If you want to leverage the out-of-the-box chat bar in the My Activity or Catalog pages) Enable BMC HelixGPT | EnableBMC HelixGPTanswer retrieval for end users. | |
5 | (If you want to enable BMC HelixGPTin studio pages) Configure the supported page components to work with BMC HelixGPT | Although this step is optional, we encourage you to enable BMC HelixGPT answer retrieval in studio pages by configuring the search component or a Search block in the Content Block component. Tip: You can configure single or multiple chat bars on the studio page. |
FAQ
For more frequently asked questions, see FAQ in the BMC HelixGPT documentation