Virtual Agent
Sera AI's conversational interface for end users on the Self Service portal.
The Virtual Agent is Sera AI's conversational interface for end users. Available on the Self Service portal, it assists users by surfacing knowledge, creating requests, highlighting pending tasks, and providing information about degraded services. The Virtual Agent is designed with a bias toward action: when a user describes what they need, it moves toward resolution rather than asking unnecessary follow-up questions.
Setting up the Virtual Agent
To enable the Virtual Agent:
- Navigate to the Account Settings section of the Settings console and ensure the Xurrent +AI toggle is enabled.
- Navigate to the Self Service Settings section and toggle the Virtual Agent on.
- In the Self Service Design section, select which self-service portals should offer the Virtual Agent.
Organizations that use the support chat functionality can continue to do so alongside the Virtual Agent.
How the Virtual Agent works
The Virtual Agent appears in the bottom-right corner of the Self Service portal. When a user types a message, Sera AI processes the request and determines the best course of action.
Finding request templates. When a user describes a need (for example, "I need a new docking station"), the Virtual Agent searches the organization's request templates to find a match. If a relevant template exists, the Virtual Agent identifies it and presents a request card with fields pre-populated from the conversation. The user can then review and submit the request. Because request templates are often linked to workflows, getting the right template from the start kicks off processes in the background without a human in the loop, leading to quicker resolution and less manual effort.
Handling requests without a matching template. When the Virtual Agent cannot find a matching request template (for example, for a brand-new service or feature), it lets the user know and offers to create a generic request instead. The user can submit this generic request, and Sera AI's auto-classification handles routing it to the right team in the background. Within seconds of submission, the request is assigned both a category and a service instance.
Pre-populating custom fields. The Virtual Agent can reference any custom fields that have been added to a given request template and pre-populate information deemed relevant and matching. The matching threshold is tempered to avoid hallucination, so custom fields may not always be filled in, particularly if the user's description was brief. If custom fields are marked as required in the request template, the user is redirected to the request page to complete those fields before submission. This avoids scenarios where a specialist has to go back to the requester for missing information.
Surfacing knowledge articles. When a user asks a question or describes a problem, the Virtual Agent searches the knowledge base and returns relevant information as concise summaries, helping users find answers without submitting a request.
Highlighting pending tasks. If a user has tasks in their inbox, the Virtual Agent summarizes them and suggests the user take action. This includes completed requests awaiting a satisfaction rating, requests with status "Waiting for Customer" that need input, and approval tasks.
Detecting degraded services. The Virtual Agent has access to the service catalog and can see if any services are currently degraded. When a user contacts the help desk about an ongoing problem, the Virtual Agent can inform them of known service issues.
Respecting support domain privacy. For organizations with strong privacy enabled, the Virtual Agent asks the user to confirm the correct support domain before routing a request. This prevents sensitive requests from being sent to the wrong support domain (for example, an HR issue being routed to Facilities).
Chat history
Virtual Agent conversations are saved in My Chats within the self-service portal. Users can review past conversations at any time. The conversation content is also captured in the request notes when a request is created, which helps specialists who pick up the request understand the full context.
Relationship with auto-classification
The Virtual Agent and AI request classification work as a connected system. When the Virtual Agent creates a request (either via a matched template or as a generic "Other" category request), the AI auto-classifier determines the appropriate Category, Impact, Service Instance, and Team. This combination of capabilities means the Virtual Agent can create well-routed requests without requiring users to navigate the service catalog manually.
The knowledge flywheel
The Virtual Agent is part of a broader knowledge management cycle. Knowledge articles enable ticket deflection through the Virtual Agent. When tickets are resolved, specialists can create new knowledge articles from the resolution. Those articles then become available for future Virtual Agent conversations. Every resolution today becomes an investment in preventing tomorrow's tickets. This flywheel accelerates over time as the knowledge base grows and improves.
Optimizing your environment for the Virtual Agent
The out-of-the-box Virtual Agent can meet needs right away for organizations with straightforward environments. For more complex scenarios, Xurrent works with customers to optimize areas such as knowledge article content, keyword coverage, and request template configuration to ensure the Virtual Agent addresses the scenarios that matter most.