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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Procurement Optimisation Agent

Procurement Optimisation Agent

Author: Venkata Sudhakar

Procurement decisions involve balancing multiple competing factors: unit price, lead time, minimum order quantities, freight costs, and supplier reliability. Manually evaluating these trade-offs across dozens of SKUs and multiple suppliers is time-consuming and error-prone.

A Procurement Optimisation Agent uses Gemini AI to evaluate purchase requisitions against available supplier quotes and automatically select the best sourcing option. It factors in total landed cost, not just unit price, giving procurement teams accurate cost comparisons.

The below example shows how ShopMax India optimises purchase orders for its quarterly electronics procurement across competing supplier quotes.

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

# Purchase requisitions and supplier quotes
REQUISITIONS = [
    {
        "sku": "Samsung Galaxy A55",
        "qty_needed": 300,
        "quotes": [
            {"supplier": "TechParts Mumbai", "unit_price": 18500, "moq": 100, "lead_days": 3, "freight_per_unit": 50},
            {"supplier": "ElecComp Shenzhen", "unit_price": 17200, "moq": 200, "lead_days": 18, "freight_per_unit": 350},
            {"supplier": "QuickDeal Delhi", "unit_price": 18800, "moq": 50, "lead_days": 2, "freight_per_unit": 80},
        ]
    },
    {
        "sku": "Boat Airdopes 141",
        "qty_needed": 500,
        "quotes": [
            {"supplier": "TechParts Mumbai", "unit_price": 1250, "moq": 200, "lead_days": 4, "freight_per_unit": 20},
            {"supplier": "SoundGear Pune", "unit_price": 1180, "moq": 500, "lead_days": 7, "freight_per_unit": 25},
        ]
    },
]

def optimise_procurement(req_json: str) -> str:
    """Select best supplier for each requisition based on total landed cost."""
    requisitions = json.loads(req_json)
    results = []

for req in requisitions:
        sku = req["sku"]
        qty = req["qty_needed"]
        best = None
        best_cost = float("inf")
        options = []

for q in req["quotes"]:
            # Check MOQ feasibility
            order_qty = max(qty, q["moq"])
            total_cost = (q["unit_price"] + q["freight_per_unit"]) * order_qty
            landed_unit = q["unit_price"] + q["freight_per_unit"]
            options.append({
                "supplier": q["supplier"],
                "order_qty": order_qty,
                "landed_unit_price": landed_unit,
                "total_cost": total_cost,
                "lead_days": q["lead_days"],
                "moq_met": qty >= q["moq"],
            })
            if total_cost < best_cost:
                best_cost = total_cost
                best = q["supplier"]

options.sort(key=lambda x: x["total_cost"])
        results.append({"sku": sku, "qty_needed": qty, "recommendation": best, "options": options})

return json.dumps(results)

procure_tool = FunctionTool(func=optimise_procurement)

AGENT_INSTRUCTION = """
You are a Procurement Optimisation Agent for ShopMax India.
When given purchase requisitions with supplier quotes:
1. Call optimise_procurement with the requisitions JSON
2. For each SKU, show a comparison table of all supplier options with landed unit price and total cost
3. Clearly state the recommended supplier and why (lowest total landed cost)
4. Flag if any supplier requires ordering more than needed due to MOQ
5. Note lead time differences - if cost difference is small, prefer faster delivery
Present as a procurement decision report in Rs.
"""

procure_agent = LlmAgent(
    name="procurement_optimise_agent",
    model="gemini-2.0-flash",
    instruction=AGENT_INSTRUCTION,
    tools=[procure_tool],
)

async def main():
    session_service = InMemorySessionService()
    runner = Runner(
        agent=procure_agent,
        app_name="procurement_app",
        session_service=session_service
    )
    session = await session_service.create_session(
        app_name="procurement_app",
        user_id="procurement-mgr"
    )
    message = Content(parts=[Part(text="Optimise procurement for these requisitions: " + json.dumps(REQUISITIONS))])
    print("Procurement Optimisation Agent - ShopMax India")
    print("=" * 50)
    async for event in runner.run_async(user_id="procurement-mgr", 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,

Procurement Optimisation Agent - ShopMax India
==================================================
PROCUREMENT DECISION REPORT

SKU: Samsung Galaxy A55 (Need: 300 units)

Supplier            Order Qty  Landed Price  Total Cost    Lead
------------------  ---------  ------------  ----------    ----
ElecComp Shenzhen      300     Rs 17,550     Rs 52,65,000  18d
TechParts Mumbai       300     Rs 18,550     Rs 55,65,000   3d
QuickDeal Delhi        300     Rs 18,880     Rs 56,64,000   2d

RECOMMENDATION: TechParts Mumbai
ElecComp Shenzhen is cheapest (saves Rs 3,00,000) but 18-day lead
time is risky given current high-demand trend. TechParts delivers
in 3 days at only Rs 3,00,000 more - worth the premium.

------------------------------------------------------------
SKU: Boat Airdopes 141 (Need: 500 units)

Supplier            Order Qty  Landed Price  Total Cost    Lead
------------------  ---------  ------------  ----------    ----
SoundGear Pune         500     Rs 1,205      Rs 6,02,500    7d
TechParts Mumbai       500     Rs 1,270      Rs 6,35,000    4d

RECOMMENDATION: SoundGear Pune
Saves Rs 32,500 with MOQ exactly matching need. 7-day lead
acceptable for this SKU. MOQ of 500 is met - no excess stock.

The agent evaluated total landed cost including freight, not just unit price, and applied lead time reasoning to its recommendations. For the Samsung Galaxy A55, it recommended the mid-priced supplier over the cheapest option because the 18-day lead time was too risky given the detected demand growth trend. This kind of multi-factor optimisation is where AI adds significant value over simple lowest-price purchasing.

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