	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Leave Approval Agent

Leave Approval Agent

Author: Venkata Sudhakar

Processing leave requests manually across ShopMax India's 2,000-employee workforce creates bottlenecks for managers and HR. An intelligent leave approval agent can instantly check leave balances, flag policy violations, assess team coverage impact, and give the manager a clear approval recommendation.

This tutorial builds a Gemini ADK agent that validates a leave request against the employee's balance, checks if the team has minimum coverage during the requested period, and returns an approval recommendation with reasoning.

The below example shows a leave approval agent for ShopMax India in a business context.

import asyncio
from google.adk.agents import LlmAgent
from google.adk.tools import FunctionTool
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai import types

# Employee leave balances
EMPLOYEES = {
    "EMP-101": {"name": "Amit Sharma", "team": "Marketing", "casual_balance": 5, "sick_balance": 8},
    "EMP-102": {"name": "Sunita Rao",  "team": "Marketing", "casual_balance": 0, "sick_balance": 10},
    "EMP-103": {"name": "Ravi Kumar",  "team": "Marketing", "casual_balance": 3, "sick_balance": 6},
}

# Team leave already approved for the requested period
APPROVED_LEAVES = {
    "Marketing": ["EMP-102"]  # Sunita already on leave Apr 10-12
}

MIN_TEAM_COVERAGE = 2  # at least 2 people must be present

def evaluate_leave(employee_id: str, leave_type: str, days_requested: int) -> dict:
    emp = EMPLOYEES.get(employee_id)
    if not emp:
        return {"error": f"Employee {employee_id} not found"}
    # Check balance
    balance_key = "casual_balance" if leave_type == "casual" else "sick_balance"
    balance = emp[balance_key]
    balance_ok = balance >= days_requested
    # Check team coverage
    team = emp["team"]
    team_size = sum(1 for e in EMPLOYEES.values() if e["team"] == team)
    already_absent = len(APPROVED_LEAVES.get(team, []))
    available_after_this = team_size - already_absent - 1  # -1 for requestor
    coverage_ok = available_after_this >= MIN_TEAM_COVERAGE
    # Build recommendation
    if balance_ok and coverage_ok:
        decision = "APPROVE"
        reason = "Sufficient balance and adequate team coverage maintained."
    elif not balance_ok:
        decision = "REJECT"
        reason = f"Insufficient {leave_type} leave balance ({balance} days available, {days_requested} requested)."
    else:
        decision = "DEFER"
        reason = f"Team coverage would drop below minimum ({MIN_TEAM_COVERAGE}). {already_absent} teammate(s) already absent."
    return {
        "employee": emp["name"],
        "employee_id": employee_id,
        "leave_type": leave_type,
        "days_requested": days_requested,
        "balance_available": balance,
        "team_coverage_ok": coverage_ok,
        "recommendation": decision,
        "reason": reason,
    }

leave_tool = FunctionTool(func=evaluate_leave)

agent = LlmAgent(
    name="leave_approval_agent",
    model="gemini-2.0-flash",
    description="Evaluates employee leave requests for ShopMax India and recommends approval decisions.",
    instruction=(
        "You are a leave management agent for ShopMax India. "
        "When an employee requests leave, call evaluate_leave with employee ID, leave type, and days. "
        "Present the recommendation (APPROVE, REJECT, or DEFER) with clear reasoning. "
        "If rejected or deferred, suggest alternatives such as splitting the leave or choosing a different date."
    ),
    tools=[leave_tool],
)

async def main():
    session_service = InMemorySessionService()
    runner = Runner(agent=agent, app_name="leave_app", session_service=session_service)
    session = await session_service.create_session(app_name="leave_app", user_id="mgr1")
    msg = types.Content(role="user", parts=[types.Part(
        text="Amit Sharma (EMP-101) has requested 2 days casual leave. Should I approve?"
    )])
    async for event in runner.run_async(user_id="mgr1", session_id=session.id, new_message=msg):
        if event.is_final_response() and event.content:
            print(event.content.parts[0].text)

asyncio.run(main())

It gives the following output,

Leave Request Evaluation - ShopMax India

Employee  : Amit Sharma (EMP-101) | Team: Marketing
Request   : 2 days Casual Leave
Balance   : 5 days casual leave available

Team Coverage Check:
  Team size: 3 | Already absent: 1 (Sunita Rao)
  Remaining if approved: 2 (meets minimum of 2)

Recommendation: APPROVE
Reason: Sufficient balance and adequate team coverage maintained.

Remaining balance after approval: 3 casual leave days.

ShopMax India can expose this agent through a manager self-service portal where every leave request triggers an instant AI evaluation. Managers still make the final decision, but the agent eliminates the need to manually check spreadsheets for balances and team rosters, cutting leave processing time from hours to seconds.

Send your comments, suggestions or queries regarding this site to [email protected].