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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Gemini API for Product Description Generation

Gemini API for Product Description Generation

Author: Venkata Sudhakar

Writing high-quality product descriptions at scale is one of the most time-consuming tasks in e-commerce. The Gemini API generates professional, brand-consistent product descriptions from raw specifications, bullet points, or even product images. For a catalogue of 10,000 products, what would take a team of writers months can be completed in hours.

ShopMax India onboards 200-300 new products per week from suppliers who provide only basic specification sheets. The content team previously spent 3-4 minutes writing each description manually. With Gemini, they generate a first draft in under 5 seconds, review it in 30 seconds, and move on - reducing content creation time by 85% and allowing the catalogue team to be redeployed to higher-value work.

The below example shows how to generate multiple variations of a product description targeting different customer segments from a single specification input.

import vertexai
from vertexai.generative_models import GenerativeModel
import json

vertexai.init(project="shopmax-india", location="asia-south1")
model = GenerativeModel("gemini-2.0-flash")

def generate_product_descriptions(specs: dict) -> dict:
    prompt = f"""Generate product descriptions for ShopMax India based on these specs.
Return ONLY a JSON object with:
- title: SEO-optimised product title (max 80 chars)
- short_description: 1-sentence description for search results (max 160 chars)
- full_description: 3-paragraph description for the product page
- bullet_points: list of 5 key selling points
- seo_keywords: list of 8 relevant search keywords
- target_audience: who this product is ideal for

Tone: professional but approachable. Highlight value for Indian customers.
Mention price competitiveness where relevant.

Product specs:
{json.dumps(specs, indent=2)}

JSON:
"""

response = model.generate_content(
        prompt,
        generation_config={"temperature": 0.7, "response_mime_type": "application/json"}
    )
    return json.loads(response.text)

specs = {
    "brand": "boAt",
    "model": "Airdopes 141",
    "category": "True Wireless Earbuds",
    "battery_life": "42 hours total (6h earbuds + 36h case)",
    "connectivity": "Bluetooth 5.1",
    "water_resistance": "IPX4",
    "driver_size": "8mm",
    "price": 1299,
    "colors": ["Active Black", "Mint Green", "Cherry Red"],
    "special_features": ["BEAST Mode for gaming", "IWP instant connect"]
}

result = generate_product_descriptions(specs)
print(f"Title: {result['title']}")
print(f"Short: {result['short_description']}")
print(f"Audience: {result['target_audience']}")
print("Bullets:")
for b in result["bullet_points"]:
    print(f"  - {b}")

It gives the following output,

Title: boAt Airdopes 141 True Wireless Earbuds with 42H Playtime - IPX4
Short: boAt Airdopes 141 TWS earbuds with 42-hour battery, BEAST gaming mode,
  and IPX4 water resistance at just Rs 1,299.
Audience: Budget-conscious students and young professionals who want reliable
  daily-use earbuds with long battery life and gaming capability.
Bullets:
  - 42-hour total playtime - never run out of music on the go
  - BEAST Mode reduces latency for lag-free mobile gaming
  - IPX4 water resistance handles sweat and light rain
  - Instant connect with IWP technology - earbuds connect in seconds
  - Bluetooth 5.1 for stable, stutter-free wireless audio

The below example shows how to batch-generate descriptions for an entire supplier product list and export to CSV.

import csv
import time

def process_supplier_catalogue(products: list, output_file: str):
    results = []

for i, product in enumerate(products):
        print(f"Processing {i+1}/{len(products)}: {product.get('model', 'Unknown')}")
        desc = generate_product_descriptions(product)
        results.append({
            "sku": product.get("sku", f"SKU-{i+1}"),
            "brand": product.get("brand"),
            "model": product.get("model"),
            "title": desc["title"],
            "short_description": desc["short_description"],
            "full_description": desc["full_description"],
            "bullet_points": " | ".join(desc["bullet_points"]),
            "seo_keywords": ", ".join(desc["seo_keywords"])
        })
        time.sleep(0.5)  # Rate limit courtesy delay

with open(output_file, "w", newline="", encoding="utf-8") as f:
        writer = csv.DictWriter(f, fieldnames=results[0].keys())
        writer.writeheader()
        writer.writerows(results)

print(f"Exported {len(results)} descriptions to {output_file}")
    return results

# Process a batch of 3 products from boAt supplier sheet
supplier_products = [
    {"sku": "BOAT-141", "brand": "boAt", "model": "Airdopes 141", "price": 1299},
    {"sku": "BOAT-181", "brand": "boAt", "model": "Rockerz 450", "price": 1499},
    {"sku": "BOAT-255", "brand": "boAt", "model": "Stone 352", "price": 1999}
]

process_supplier_catalogue(supplier_products, "boat_descriptions.csv")

It gives the following output,

Processing 1/3: Airdopes 141
Processing 2/3: Rockerz 450
Processing 3/3: Stone 352
Exported 3 descriptions to boat_descriptions.csv

The temperature of 0.7 gives descriptions that are creative and varied rather than formulaic, while still being accurate to the specs. ShopMax India runs this pipeline for every new supplier onboarding, generating a complete set of titles, descriptions, bullet points, and SEO keywords before the products go live. The output CSV feeds directly into the product information management system, eliminating the manual copy-paste step that was previously the content team bottleneck.

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