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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Vendor Performance Scoring Agent

Vendor Performance Scoring Agent

Author: Venkata Sudhakar

ShopMax India works with dozens of electronics vendors. Some deliver on time with zero defects; others cause constant stock shortages and quality complaints. Without a systematic scoring framework, procurement decisions rely on relationships rather than data. A vendor performance agent that scores each supplier on delivery, defect rate, invoice accuracy, and price competitiveness gives the procurement team objective data to negotiate better terms and drop underperforming vendors.

This tutorial builds a Vendor Performance Scoring Agent using ADK and Gemini. The agent aggregates quarterly data for each vendor, computes a weighted performance score across four dimensions, and produces a ranked scorecard with recommended actions for each vendor relationship.

The below example shows a quarterly vendor scorecard for ShopMax India suppliers.

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

# --- Vendor quarterly performance data ---
VENDOR_DATA = {
    "Samsung India Pvt Ltd": {
        "on_time_delivery_pct": 94,
        "defect_rate_pct": 0.8,
        "invoice_accuracy_pct": 97,
        "price_vs_market_pct": -3,  # 3% below market = competitive
        "total_orders_q": 12,
        "category": "tv",
    },
    "Sony India Ltd": {
        "on_time_delivery_pct": 88,
        "defect_rate_pct": 1.5,
        "invoice_accuracy_pct": 100,
        "price_vs_market_pct": 2,   # 2% above market
        "total_orders_q": 8,
        "category": "audio",
    },
    "Apple India Distributor": {
        "on_time_delivery_pct": 78,
        "defect_rate_pct": 0.3,
        "invoice_accuracy_pct": 95,
        "price_vs_market_pct": 5,   # 5% above market
        "total_orders_q": 6,
        "category": "tablet",
    },
}

# Scoring weights
WEIGHTS = {
    "delivery": 0.35,
    "quality": 0.30,
    "invoice": 0.20,
    "price": 0.15,
}

# --- Tool: score a single vendor ---
def score_vendor(vendor_name: str) -> dict:
    """Compute weighted performance score for a vendor.
    Args:
        vendor_name: Full vendor name as in the database
    Returns:
        Dimension scores, weighted total, and performance tier
    """
    data = VENDOR_DATA.get(vendor_name)
    if not data:
        return {"error": f"Vendor not found: {vendor_name}"}
    # Delivery score (0-100)
    delivery_score = data["on_time_delivery_pct"]
    # Quality score: 0% defect = 100, 5% defect = 0
    quality_score = max(0, 100 - data["defect_rate_pct"] * 20)
    # Invoice accuracy score
    invoice_score = data["invoice_accuracy_pct"]
    # Price score: at market = 70, below market = up to 100, above market = down to 0
    price_score = min(100, max(0, 70 - data["price_vs_market_pct"] * 6))
    weighted = (
        delivery_score * WEIGHTS["delivery"] +
        quality_score * WEIGHTS["quality"] +
        invoice_score * WEIGHTS["invoice"] +
        price_score * WEIGHTS["price"]
    )
    total = round(weighted, 1)
    if total >= 85:
        tier = "PREFERRED"
    elif total >= 70:
        tier = "APPROVED"
    elif total >= 55:
        tier = "WATCH"
    else:
        tier = "AT_RISK"
    return {
        "vendor": vendor_name,
        "delivery_score": delivery_score,
        "quality_score": round(quality_score, 1),
        "invoice_score": invoice_score,
        "price_score": round(price_score, 1),
        "total_score": total,
        "tier": tier,
        "category": data["category"],
    }

# --- Tool: score all vendors ---
def score_all_vendors() -> dict:
    """Produce full vendor scorecard for all tracked suppliers.
    Returns:
        Ranked vendor list by score
    """
    results = [score_vendor(v) for v in VENDOR_DATA]
    results.sort(key=lambda x: x.get("total_score", 0), reverse=True)
    return {"total_vendors": len(results), "scorecards": results}

# --- Agent setup ---
vendor_tool = FunctionTool(func=score_vendor)
all_tool = FunctionTool(func=score_all_vendors)

agent = LlmAgent(
    name="vendor_scoring_agent",
    model="gemini-2.0-flash",
    instruction="""You are a vendor performance analyst for ShopMax India.
Call score_all_vendors to generate the full quarterly scorecard.
Present results as a ranked table: Rank | Vendor | Delivery | Quality | Invoice | Price | Total | Tier.
For each tier provide a recommended action:
  PREFERRED: Increase order volumes and negotiate long-term contracts.
  APPROVED: Maintain current relationship, monitor for improvement.
  WATCH: Issue performance warning, set 90-day improvement target.
  AT_RISK: Consider alternate suppliers, reduce exposure.
Highlight the single weakest dimension for each WATCH or AT_RISK vendor.""",
    tools=[vendor_tool, all_tool],
)

session_service = InMemorySessionService()
runner = Runner(agent=agent, app_name="vendor_scoring", session_service=session_service)

import asyncio

async def run_scoring(query: str) -> str:
    session = await session_service.create_session(app_name="vendor_scoring", user_id="procurement")
    from google.genai.types import Content, Part
    msg = Content(role="user", parts=[Part(text=query)])
    result = ""
    async for event in runner.run_async(user_id="procurement", session_id=session.id, new_message=msg):
        if event.is_final_response() and event.content:
            result = event.content.parts[0].text
    return result

response = asyncio.run(run_scoring("Generate the Q1 vendor scorecard and tell me which vendors need attention."))
print(response)

It gives the following output,

Vendor Performance Scorecard - ShopMax India Q1

Rank | Vendor                  | Delivery | Quality | Invoice | Price | TOTAL | Tier
-----|-------------------------|----------|---------|---------|-------|-------|----------
  1  | Samsung India Pvt Ltd   |   94.0   |  84.0   |  97.0   | 88.0  | 91.4  | PREFERRED
  2  | Sony India Ltd          |   88.0   |  70.0   | 100.0   | 58.0  | 80.4  | APPROVED
  3  | Apple India Distributor |   78.0   |  94.0   |  95.0   | 40.0  | 78.2  | APPROVED

Recommended Actions:

Samsung India (PREFERRED): Increase TV order volumes and open contract
negotiations for FY26. Strong across all dimensions.

Sony India (APPROVED): Monitor pricing - 2% above market is manageable
but should be addressed in next contract renewal. Quality defect rate
of 1.5% is the weakest dimension - request quality audit.

Apple India Distributor (APPROVED): Delivery at 78% is the primary concern
- 22% of orders arriving late creates stockout risk for high-demand iPads.
Raise formal delivery SLA discussion before next quarter order.

Schedule this agent to run at the end of each quarter using Cloud Scheduler. Pull vendor data directly from your ERP or BigQuery orders table. Store scorecards in Firestore and send the report to the procurement team via email using Cloud Functions. Track score trends across quarters to identify improving or declining vendor relationships.

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