HR Companion Agent

Agent Description:

The HR Companion Agent is an intelligent assistant that supports employees with instant answers to HR policy questions, performs basic HR transactions, and ensures complex issues are routed to human HR staff. It acts as a first-line digital interface for everyday HR needs.
Purpose and Components
  • Purpose: To automate employee support for common HR policy queries and routine actions, while escalating sensitive or unresolved issues to human HR personnel.
  • Components:
    • A policy retrieval agent for answering HR questions using internal knowledge.
    • An operations agent to perform simple HRIS interactions (example, leave balance, contact update).
    • An escalation agent to flag complex or unresolved issues for human review.
Supported Capabilities
  • Answering employee queries related to HR policies (leave, benefits, conduct, etc.)
  • Performing predefined HR operations like checking leave balance or submitting leave requests.
  • Automatically routing complex, sensitive, or unresolved queries to the HR team.
  • Maintaining polite, professional interaction quality across all scenarios.
LLM Used
  • Google Vertex
    Note: To learn more about the LLM and to modify its behavior, refer to the Configuring LLM settings section.

Sub-Agents

1. Policy Knowledge Agent

  • Role: Query Responder.
  • Scope: Retrieve policy-related answers from an HR knowledge base document.
  • Description: This subagent receives employee questions from structured or unstructured input and determines the policy category (for example, leave, benefits, dress code). It then performs semantic retrieval over the HR policy document and returns a concise, professional response to the employee’s question. Example topics include work hours, maternity leave, holiday schedules, and conduct rules.
  • Tools used: MCP Client Tool

2. HR Operation Agent

  • Role: HR Operator.
  • Scope: Handle HR tasks like checking balances, updating contact info, and applying for leave.
  • Description: This subagent identifies when a user’s query involves a transactional HR request (e.g., “What’s my leave balance?”). It simulates interaction with backend systems (via mock API or GET Request Tool), processes the action, and returns a confirmation message or data view. It supports predefined HR operations including:
    • Viewing leave balances
    • Applying for leave
    • Updating emergency contact
    • Viewing holiday calendar
  • Tools used: GET Request Tool

3. Escalation and Ticketing Agent

  • Role: Issue Escalator.
  • Scope: Review query history and decide whether escalation is needed.
  • Description: This final subagent receives the conversation flow and determines if the employee’s issue has been addressed. If a previous subagent provided a valid, complete answer, it confirms that response to the user. If the issue is too complex, sensitive (for example, harassment report), or unanswered, it triggers a ticket and informs the user that a human HR representative will follow up within 3–5 business days. Escalation is only used when necessary.
Tools Used
  • MCP Client Tool (for knowledge retrieval from policy documents)
  • GET Request Tool (for HRIS simulation tasks)
Note: For details on modifying the Tools, refer Tools Library section.
Agent Workflow Behavior Summary
  • The Policy Knowledge Agent receives the query and retrieves an answer from the HR policy document.
  • If the query is action-oriented, the HROperationAgent performs the corresponding HR task and returns confirmation.
  • The Escalation and Ticketing Agent determines whether the previous subagents resolved the query. If not, or if the query is sensitive, it raises a ticket and informs the user of next steps.
Sample Questions:
  • How many annual leave days do I get?
  • When is the salary typically credited?
  • Where can I find the list of company holidays?