Using Employee Navigator to find AI-generated answers

Text

Images

Users can have a real-time, multi-turn voice conversation with Employee Navigator by clicking the Audio Wave button. The assistant listens and replies in natural language by using voice. All spoken input is transcribed and saved in the session history until the user logs out or starts a new conversation.

Employee Navigator answers queries by using various sources, including knowledge articles, service requests, approvals, to-dos, and user profile data. It only uses content that the user is permitted to view, based on their access rights and filtering rules.

To begin a conversation, users must grant microphone access and consent to share audio. Browsers remember microphone permissions, but consent is requested at the start of each session.

During the session, users can perform the following actions:

• Control how Employee Navigator listens to your input by muting or unmuting the microphone:
  • When muted, Employee Navigator doesn't respond to voice input but accepts and replies to text.
  • If unmuted but silent, Employee Navigator continues listening and waits for the next input.
• Mix voice and text input at any time.
  WarningUsers can only send text messages when the microphone is muted. If the microphone is active, the text input option will remain disabled.
• Interrupt responses by speaking.
• End the session by clicking the End session button.

Voice input (transcripted to text)

Users can share their screen with Employee Navigator to visually present their task and receive real-time,context-aware assistance. When screen sharing begins, voice-to-voice streaming is automatically activated, enabling users to speak naturally while working. The voice experience functions as described in the Voice-to-voice streaming section.

Employee Navigator observes the shared screen without interacting with it and uses visual context along with spoken or typed input to provide voice-based guidance.

Employee Navigator answers queries by referencing various sources, including knowledge articles, service requests, approvals, to-dos, and user profile data. It only uses the content that the user is permitted to view, based on their access rights and filtering rules.

To begin screen sharing, users click the Screen Share button in the chat window and select the screen, window, or browser tab to share. If microphone access hasn’t been granted, the browser prompts for permission to access the microphone.

A consent dialog box appears when screen sharing starts for the first time in a session and will appear again after the user logs out and logs back in or refreshes the browser page.

During the session, users can perform the following actions:

• Pause or resume screen sharing by clicking the Screen Share button. Voice interaction continues even when screen sharing is paused.
• Mute and unmute the microphone.
• Mix voice and text input at any time.
• Interrupt responses by speaking or clicking the Stop button.
• End the session by clicking the End session button. Ending screen sharing also ends the voice session.

From mobile devices

From end-user console

From Studio pages

From the Catalog and My Activity pages

Users can access Employee Navigator chat within the Microsoft Teams interface to find answers to workplace queries or manage service requests. The chat supports natural language input and provides structured responses, similar to the experience in the end-user console. However, Employee Navigator responses in Microsoft Teams appear in a bulleted format instead of tabular layouts used in the native console.

This integration enables users to ask queries directly within Microsoft Teams and move towards resolution by leveraging agent-based configuration, ensuring smooth experience. Any links included in the responses guide users to the native application for further action. For more information, see Configuring Employee Navigator in the Microsoft Teams application.

Employee Navigator uses AI to understand user intent and suggests a list of relevant catalog services as links in the chat. The links provide quick access to request submission pages for the suggested services that cannot be fulfilled via chat. The suggestions are generated based on the user’s query and user entitlements. This capability improves the self-service experience, reduces support overhead, and helps users submit service requests more efficiently.

The following agentic skills are supported for this capability:

• DWP Employee Navigator Agent - V1
• DWP Employee Navigator Agent - V1 - Cards

The following model is supported:

• OpenAI Azure GPT-4.1

End users can submit service requests via the Employee Navigator chat through an interactive conversational experience. When a user queries about submitting a request, Employee Navigator interprets their query, identifies, and presents a list of the most relevant catalog services. If the user selects a chat-supported catalog service, it guides them through a conversational questionnaire.

Users provide input for the questionnaire by using an interactive UI within a series of question components in the Employee Navigator chat. Depending on the question type, the question component presents an interactive UI for users to provide their responses, such as an embedded text area, a dropdown menu, a typeahead lookup field, a date/time picker, or an attachment uploader. The question component includes the question label and, a description, if defined during the questionnaire configuration. If a question was configured with a default value, the question component will be prefilled with it.

Each question component also provides a Skip and Submit response button, with the following behavior:

For Required questions:

• The Skip button is available only if a default value is configured.
• The Submit response button remains disabled until an input is provided.

For Optional questions:

• The Skip button is always available, regardless of default values.
• The Submit response button remains disabled until an input is provided.

For more information about the chat-enabled service requests, see Using BMC HelixGPT to make services requestable via chat.

End users can connect to a live agent directly from within the Employee Navigator chat if they do not receive relevant answers to knowledge article-related queries from Employee Navigator and need human assistance.

End users can choose to connect with a live agent in the following situations:

• No results found
• Failed service requests
• Failed router classification
• System errors
• General chatbot assistant failures
• Human assistance is preferred

To learn about the agent-based configuration, see Configuring the Supervisor agent for Employee Navigator.

End users can add a comment to their active service requests directly from BMC HelixGPT chat in BMC Helix Digital Workplace. This is achieved by using the following prompt:

Add a comment "abc" to request <id or title> 

The comment input supports RTF within the chat interface, enabling users to format their text (such as bold, italics, or lists) for better clarity and expression. However, not all service requests support RTF comments. Administrators perform configuration to accept RTF comments for the RTF editor to appear.

Note:

To use the RTF editor and see the Add Comment/Attachment button, users must avoid including the actual comment in the initial prompt.

• If the user says: “Add comment 'abc' to request <ID>”, the system adds the comment directly without opening the RTF editor.
• If the user says: “Add a comment to request <ID>”, the system prompts the user to enter the comment using the RTF editor or button.

Important:

• ​​​​​You can add only one comment per service request at a time. The comment must be within the maximum limit of 2048 characters.

• The system hides comments that contain only attachments and no text when you request the list of comments; for example, when you use the prompt:
  List the comments from request <ID>

• The RTF editor in BMC HelixGPT chat does not display the 'maximum image count' message. This message is visible in the BMC Helix Digital Workplace comment section but not in the BMC HelixGPT chat interface.
• After the user adds a comment successfully, the Add comment button is disabled, and the submitted comment appears underneath the input area with the user’s avatar and the date the comment was added. If the chat displays multiple buttons for comments or attachments, the system automatically disables previously displayed buttons and enables only the latest one.

Best practice: If the iPhone wrapper app crashes while using the RTF editor in chat, we recommend clearing the app cache. This issue is usually related to memory or cache overload.

This capability only works with Agent skills, such as DWP Employee Navigator Agent - V1 or DWP Employee Navigator Agent - V1 - cards. These agent skills are preconfigured with the associated prompts. Administrators must configure these agent skills for BMC HelixGPT chat or Employee Navigator. 

For more information, see Prebuilt skills in BMC Helix Digital Workplace.

You can optionally include one or more attachments along with the comment, depending on the attachment settings configured in the admin console.

For details about allowed file types, size limits, and configuration steps for attachments in comments and chat, see Updating the attachment settings and URI schemes.

End users can cancel a service request directly from the Employee Navigator chat in BMC Helix Digital Workplace. This is achieved by using the following prompt:

Cancel request <id or title>

When this prompt is entered, a summary of the request is shown, and the user is asked for a final confirmation of the cancellation. Upon confirmation, the request is canceled, and the user is shown the final summary with the updated status that reads canceled.

This capability only works with Agent skills, such as DWP Employee Navigator Agent - V1 or DWP Employee Navigator Agent - V1 - cards. These agent skills are preconfigured with the associated prompts. Administrators must configure these agent skills for BMC HelixGPT chat or Employee Navigator.

For more information, see Prebuilt skills in BMC Helix Digital Workplace.

Allows users to view a condensed summary of all comments on a service request. By default, it summarizes all comments approximately in 100 words. Users can specify a word count or ask to view all or selected comments.

If users do not specify which comments to summarize, the system includes the top 10 comments by default.

Important:

• The OpenAI Azure GPT-4.1 and OpenAI Azure GPT-4o models support this feature.
• Comment summarization supports only text-based content. Images, videos, and other non-text formats are not summarized.
• You can generate summaries for requests in any status (active, closed, or canceled), as long as they contain comments.
• Fulfillment-type ticket comment summarization is not supported.
• Adding comments to or cancelling a fulfillment request is not supported.
• Very large comments may exceed the language model's token limit and cause summarization to fail.
• To-do items are not supported for adding comments or attachments via Employee Navigator.

For more information about how to modify the summarization default configuration, see Configuring Employee Navigator in the end-user console.

Approvers can approve or reject an approval request directly from the Employee Navigator by using the request ID or name.

To approve a request, users can enter a prompt such as:

Can you approve <request ID> for me?

To reject a request, users can enter a prompt such as:

Can you reject <request ID> for me?

When this prompt is submitted, the user is asked to provide a justification. The user can enter a reason, such as:

I am approving this request now.

or

You don't need this.

After the justification is submitted, the request is approved or rejected based on the prompt provided, and the updated details are displayed. Alternatively, users can include the justification in the original prompt. 

End users can reassign an approval request to another user directly from the Employee Navigator.

To reassign a request, users can enter a prompt such as:

Can you reassign <request ID> to <user name or user ID>?

If the specified user name or user ID matches multiple users, the Employee Navigator displays all of them in a table view. The user is then prompted to select the intended recipient for the reassignment.

After a user is selected, the Employee Navigator asks for final confirmation. Once confirmed, the approval request is reassigned to the selected user, and the updated request details are displayed.

Once the request is reassigned, it is updated with the new approver name. Additionally, the request disappears from the list of your active events and you cannot access that request.

End users can place an approval request on hold directly from the Employee Navigator by using the request ID or name.

To put a request on hold, users can enter a prompt such as:

Put <request ID> on hold.

If the justification is not included in the original prompt, the Employee Navigator prompts the user to provide one. The user can enter a justification, such as:

The request is pending additional information.

After the justification is submitted, the request is placed on hold, and the updated request details are displayed.

Service requests often generate fulfillment tickets such as work orders, incidents, cases, or change requests, each with its own unique ID.

End Users can search for the associated service request in the Employee Navigator using any of these fulfillment IDs. End users can also query details like the status of a service request by referencing any fulfillment ID in their prompts, for example:

Show me the status of the service request with work order <work order ID>

For more information, see Configuring fulfillment details display in service requests.

Employee Navigator offers a prebuilt welcome message and starter prompts for the Employee Navigator for Helix skill. 

When users start a new Employee Navigator chat that uses this skill, they can use the starter prompts to perform the following tasks:

• Report an issue
• View requests
• Check approvals
• Access To‑dos 
• Add a comment to a request
• Summarize the comments on a request
• Ask Employee Navigator to describe what it can do

The starter prompts help complete tasks quickly, reduce unnecessary conversational steps, and support faster adoption.

Administrators can also configure a custom welcome message and starter prompts for any custom skill so users can perform actions tailored to that skill. For more information, see Configuring Employee Navigator in the end-user console.

Knowledge articles

When a user asks questions in the Employee Navigator UI, it responds by providing answers sourced from the available knowledge. It can also show images from the knowledge articles to enrich the information.

Service requests and approvals

To-dos

People attributes

Service health items

Streaming

Grid view

(Controlled availability customers from version 23.3.04 and later) Card view

Click Copy  to copy the response generated by Employee Navigator to the clipboard. The copy option is available for both agent-based and prompt-based skills. You can paste this content into other applications that support HTML-formatted content, such as Word or email.

The copied content retains formatting and includes images and in-line or source links IF they are present.

Important:

• The copied content can expose sensitive data, such as passwords as they cannot be masked.
• Embedded images from a protected environment that need access authorization appear as broken links when they are pasted outside that environment.
• Videos cannot be copied by using this option.
• This option is not available when:
  • Users connect to a live agent in the Employee Navigator chat.
  • Service and approval requests are displayed as card views.

Employee Navigator uses GenAI to interpret user questions and route them to the most appropriate resolution path.

• Asks users follow-up questions to clarify their ambiguous inputs before proceeding.
• Provides relevant knowledge or troubleshooting guidance based on user intent retrieved from the knowledge articles.
• Walks users through a service questionnaire to submit a request via chat for supported catalog services.
• Suggests catalog service links when chat-based submission isn't supported or no knowledge is retrieved.
• Connects users to a live agent or provides a generic request link when no resolution path is available.

This behavior is enabled by configuring the Employee Navigator Supervisor Agent. For more information, see Configuring the Supervisor agent for Employee Navigator.

Localization

Feedback to responses

Reliability