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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Extended Warranty Upsell Agent

Extended Warranty Upsell Agent

Author: Venkata Sudhakar

Extended warranties are high-margin products for ShopMax India, but the window to sell them is narrow - most customers decide within the first 30 days of purchase. An upsell agent that identifies which customers bought high-value electronics, checks whether they already have warranty coverage, and triggers a targeted offer at day 14 post-purchase can significantly boost warranty attachment rates without annoying customers who have already opted in.

This tutorial builds an Extended Warranty Upsell Agent using ADK and Gemini. The agent scans recent purchases, scores each customer's upsell propensity based on product value, category, and purchase history, and generates a personalised warranty offer message.

The below example shows the warranty upsell workflow for ShopMax India electronics customers.

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

# --- Warranty plans available ---
WARRANTY_PLANS = {
    "electronics_1yr": {"name": "1-Year Extended Warranty", "price": 1499, "covers": "Manufacturing defects + accidental damage"},
    "electronics_2yr": {"name": "2-Year Extended Warranty", "price": 2499, "covers": "Manufacturing defects + accidental damage + free service visits"},
    "premium_3yr": {"name": "3-Year Premium Care", "price": 3999, "covers": "All risks including liquid damage + priority service in 48 hours"},
}

# --- Recent purchases (day 14, warranty window open) ---
PURCHASES = [
    {"order_id": "ORD-6001", "customer_name": "Suresh Iyer", "city": "Chennai",
     "product": "Samsung 55-inch 4K TV", "product_value": 62000,
     "category": "tv", "days_since_purchase": 14, "has_warranty": False,
     "previous_warranty_purchases": 0},
    {"order_id": "ORD-6002", "customer_name": "Anita Desai", "city": "Mumbai",
     "product": "Apple iPad Air 11-inch", "product_value": 74000,
     "category": "tablet", "days_since_purchase": 14, "has_warranty": True,
     "previous_warranty_purchases": 2},
    {"order_id": "ORD-6003", "customer_name": "Kiran Rao", "city": "Bangalore",
     "product": "Sony WH-1000XM5 Headphones", "product_value": 29000,
     "category": "audio", "days_since_purchase": 14, "has_warranty": False,
     "previous_warranty_purchases": 1},
]

# --- Tool: get upsell candidates ---
def get_upsell_candidates() -> dict:
    """Fetch customers at day 14 post-purchase who do not yet have warranty coverage.
    Returns:
        List of customers eligible for warranty upsell
    """
    candidates = [p for p in PURCHASES if not p["has_warranty"]]
    return {"total": len(candidates), "candidates": candidates}

# --- Tool: score upsell propensity and recommend plan ---
def score_warranty_upsell(order_id: str) -> dict:
    """Score a customer for warranty upsell propensity and recommend the best plan.
    Args:
        order_id: Order identifier
    Returns:
        Propensity score, recommended plan, and offer price
    """
    purchase = next((p for p in PURCHASES if p["order_id"] == order_id), None)
    if not purchase:
        return {"error": f"Order not found: {order_id}"}
    score = 0
    # High-value product
    if purchase["product_value"] >= 50000:
        score += 40
    elif purchase["product_value"] >= 20000:
        score += 25
    else:
        score += 10
    # Previous warranty buyer
    if purchase["previous_warranty_purchases"] > 0:
        score += 30
    # High-risk categories (more likely to need repair)
    if purchase["category"] in ["tv", "laptop", "tablet"]:
        score += 20
    else:
        score += 10
    # Recommend plan based on product value
    if purchase["product_value"] >= 50000:
        plan = WARRANTY_PLANS["premium_3yr"]
    elif purchase["product_value"] >= 25000:
        plan = WARRANTY_PLANS["electronics_2yr"]
    else:
        plan = WARRANTY_PLANS["electronics_1yr"]
    return {
        "order_id": order_id,
        "customer_name": purchase["customer_name"],
        "product": purchase["product"],
        "product_value": purchase["product_value"],
        "propensity_score": score,
        "recommended_plan": plan["name"],
        "plan_price": plan["price"],
        "coverage": plan["covers"],
    }

# --- Agent setup ---
candidates_tool = FunctionTool(func=get_upsell_candidates)
score_tool = FunctionTool(func=score_warranty_upsell)

agent = LlmAgent(
    name="warranty_upsell_agent",
    model="gemini-2.0-flash",
    instruction="""You are an extended warranty upsell agent for ShopMax India.
1. Call get_upsell_candidates to find customers without warranty coverage.
2. For each candidate, call score_warranty_upsell to get propensity score and plan.
3. For HIGH propensity (score >= 60): generate a personalised, value-focused message
   explaining what the warranty covers and the peace-of-mind it provides.
4. For MEDIUM propensity (score 30-59): keep the message shorter and more casual.
5. Skip LOW propensity customers (score < 30).
Always mention the specific product, warranty price in Rs, and coverage highlights.
Do not sound pushy - frame it as helpful protection for their investment.""",
    tools=[candidates_tool, score_tool],
)

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

import asyncio

async def run_upsell(query: str) -> str:
    session = await session_service.create_session(app_name="warranty_agent", user_id="sales_team")
    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="sales_team", 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_upsell("Run the warranty upsell campaign for all day-14 customers without coverage."))
print(response)

It gives the following output,

Warranty Upsell Campaign - ShopMax India
2 candidates identified (1 skipped - already has warranty)

--- ORD-6001 | Suresh Iyer, Chennai | HIGH PROPENSITY (score: 60) ---
Product: Samsung 55-inch 4K TV (Rs 62,000)
Recommended: 3-Year Premium Care - Rs 3,999

Message: Hi Suresh, your Samsung 4K TV is now 2 weeks old - hope it is
lighting up your living room! For just Rs 3,999, our 3-Year Premium Care plan
protects it from manufacturing defects, accidental damage, and liquid damage
with priority 48-hour service. That is less than Rs 4 per day to protect
your Rs 62,000 investment. Activate before day 30: shopmax.in/warranty/ORD-6001

--- ORD-6003 | Kiran Rao, Bangalore | MEDIUM PROPENSITY (score: 35) ---
Product: Sony WH-1000XM5 Headphones (Rs 29,000)
Recommended: 2-Year Extended Warranty - Rs 2,499

Message: Hi Kiran, enjoy your Sony headphones! Consider adding our
2-Year Extended Warranty (Rs 2,499) to cover repairs and accidental damage.
Offer valid until day 30. shopmax.in/warranty/ORD-6003

Automate this agent with a Cloud Scheduler trigger on day 14 post-purchase. Connect it to your WhatsApp Business API or push notification service to deliver messages directly to the customer's device. Log each upsell attempt and outcome in BigQuery to measure attachment rate by product category and optimise the propensity scoring model over time.

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