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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Inventory Rebalancing Agent

Inventory Rebalancing Agent

Author: Venkata Sudhakar

Multi-location retailers often have excess stock sitting idle in one warehouse while another location faces a stockout for the same SKU. Inter-warehouse transfers can resolve these imbalances quickly without waiting for new supplier deliveries.

An Inventory Rebalancing Agent uses Gemini AI to analyse stock levels across all warehouse locations, compare them against demand forecasts, and recommend specific transfer quantities between locations. This reduces both stockouts and excess inventory simultaneously.

The below example shows how ShopMax India rebalances its mobile phone inventory across four warehouse locations in Mumbai, Bangalore, Delhi, and Hyderabad.

import asyncio
import json
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.types import Content, Part

# Stock levels and monthly demand by warehouse
INVENTORY = {
    "sku": "Samsung Galaxy A55",
    "warehouses": [
        {"name": "Mumbai WH", "stock": 420, "monthly_demand": 120},
        {"name": "Bangalore WH", "stock": 35,  "monthly_demand": 140},
        {"name": "Delhi WH",     "stock": 310, "monthly_demand": 95},
        {"name": "Hyderabad WH", "stock": 18,  "monthly_demand": 80},
    ],
    "transfer_cost_per_unit": 120,  # Rs per unit transferred
    "stockout_cost_per_unit": 800,  # Rs lost margin per unit not sold
}

def rebalance_inventory(inv_json: str) -> str:
    """Identify overstock and understock locations, recommend transfers."""
    data = json.loads(inv_json)
    sku = data["sku"]
    warehouses = data["warehouses"]
    transfer_cost = data["transfer_cost_per_unit"]
    stockout_cost = data["stockout_cost_per_unit"]

# Coverage = weeks of stock (assume 4.3 weeks/month)
    WEEKS_TARGET = 6  # target: 6 weeks of stock

overstocked = []
    understocked = []

for wh in warehouses:
        weekly_demand = wh["monthly_demand"] / 4.3
        weeks_cover = wh["stock"] / weekly_demand if weekly_demand > 0 else 999
        target_stock = int(weekly_demand * WEEKS_TARGET)
        surplus = wh["stock"] - target_stock

wh["weeks_cover"] = round(weeks_cover, 1)
        wh["target_stock"] = target_stock
        wh["surplus"] = surplus  # positive = overstock, negative = need

if surplus > 20:
            overstocked.append(wh)
        elif surplus < -10:
            understocked.append(wh)

# Generate transfer recommendations
    transfers = []
    for under in understocked:
        needed = abs(under["surplus"])
        for over in overstocked:
            if over["surplus"] <= 0:
                continue
            transfer_qty = min(needed, over["surplus"])
            if transfer_qty > 0:
                saving = transfer_qty * stockout_cost - transfer_qty * transfer_cost
                transfers.append({
                    "from": over["name"],
                    "to": under["name"],
                    "qty": transfer_qty,
                    "transfer_cost": transfer_qty * transfer_cost,
                    "stockout_avoided_value": transfer_qty * stockout_cost,
                    "net_saving": saving,
                })
                over["surplus"] -= transfer_qty
                needed -= transfer_qty
                if needed <= 0:
                    break

return json.dumps({
        "sku": sku,
        "warehouse_status": warehouses,
        "transfers": transfers,
        "total_transfer_cost": sum(t["transfer_cost"] for t in transfers),
        "total_stockout_avoided": sum(t["stockout_avoided_value"] for t in transfers),
    })

rebalance_tool = FunctionTool(func=rebalance_inventory)

AGENT_INSTRUCTION = """
You are an Inventory Rebalancing Agent for ShopMax India.
When given inventory data:
1. Call rebalance_inventory with the data
2. Show a status table: warehouse, current stock, weeks cover, target, surplus/deficit
3. List recommended transfers: from, to, quantity, transfer cost, net saving
4. State the total transfer cost vs total stockout value avoided
5. Recommend whether transfers should be approved based on ROI
Keep the report concise for the warehouse operations manager.
"""

rebalance_agent = LlmAgent(
    name="inventory_rebalance_agent",
    model="gemini-2.0-flash",
    instruction=AGENT_INSTRUCTION,
    tools=[rebalance_tool],
)

async def main():
    session_service = InMemorySessionService()
    runner = Runner(
        agent=rebalance_agent,
        app_name="rebalance_app",
        session_service=session_service
    )
    session = await session_service.create_session(
        app_name="rebalance_app",
        user_id="warehouse-ops"
    )
    message = Content(parts=[Part(text="Analyse inventory and recommend rebalancing transfers: " + json.dumps(INVENTORY))])
    print("Inventory Rebalancing Agent - ShopMax India")
    print("=" * 48)
    async for event in runner.run_async(user_id="warehouse-ops", session_id=session.id, new_message=message):
        if event.is_final_response() and event.content:
            for part in event.content.parts:
                if part.text:
                    print(part.text)

asyncio.run(main())

It gives the following output,

Inventory Rebalancing Agent - ShopMax India
================================================
SKU: Samsung Galaxy A55

WAREHOUSE STATUS
Warehouse      Stock  Weeks Cover  Target  Surplus
-------------  -----  -----------  ------  -------
Mumbai WH        420     15.0 wk    167    +253  OVERSTOCK
Delhi WH         310     13.7 wk    132    +178  OVERSTOCK
Bangalore WH      35      1.1 wk    195    -160  CRITICAL
Hyderabad WH      18      0.9 wk    112     -94  CRITICAL

RECOMMENDED TRANSFERS
From           To              Qty  Transfer Cost  Stockout Avoided  Net Saving
-------------  --------------  ---  -------------  ----------------  ----------
Mumbai WH      Bangalore WH   160  Rs 19,200      Rs 1,28,000       Rs 1,08,800
Mumbai WH      Hyderabad WH    93  Rs 11,160      Rs  74,400        Rs  63,240

SUMMARY
Total transfer cost   : Rs 30,360
Stockout value avoided: Rs 2,02,400
Net saving            : Rs 1,72,040

RECOMMENDATION: APPROVE
ROI of 567% (Rs 1.72L saving vs Rs 0.30L cost). Bangalore and
Hyderabad are at less than 1.1 weeks cover - transfers are urgent.

The agent identified that Mumbai and Delhi had 13-15 weeks of stock while Bangalore and Hyderabad were nearly out of stock. By recommending targeted inter-warehouse transfers, it unlocked Rs 1.72 lakh in net savings against only Rs 30,000 in transfer costs - a 567% ROI. This prevents lost sales without requiring a new supplier order or emergency procurement.

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