Automate e-commerce product listing creation from supplier catalogue images and descriptions at scale

You're managing a clothing brand with 500 SKUs across multiple suppliers. Each week, you receive updated product images and basic descriptions via email or shared spreadsheets. Right now, your team spends 40 hours writing unique descriptions, editing images to remove backgrounds, resizing them for different marketplaces, and uploading everything manually to Shopify, Amazon, and eBay. A single product might need five different image versions and three entirely different description angles depending on where it's selling. By the time you've finished listing 50 items, the season has moved on and you're already behind on next week's batch. The opportunity here is straightforward: you can automate 80 percent of this work in one weekend using a combination of AI image tools, copywriting automation, and a workflow orchestrator. The remaining 20 percent, your human review step, should take minutes instead of hours. This post walks you through wiring together AI Boost, Copy.ai, Pixelcut AI, and your orchestration platform to transform raw supplier data into polished, multi-marketplace-ready listings. The goal is zero manual handoff between tools. You upload a supplier catalogue image and a basic product description once. The workflow generates backgrounds, writes marketplace-specific copy, resizes images, and hands you a clean JSON file ready for bulk import into any storefront. For more on this, see E-commerce product description factory: generate listings.... For more on this, see Transform product images into marketplace listings with d.... For more on this, see Generate e-commerce product listings at scale with descri....

The Automated Workflow

Start by choosing your orchestration platform. For this workflow, n8n works well if you want self-hosted control and don't mind managing webhooks yourself. Zapier is faster to set up but costs more at scale. Make (Integromat) sits somewhere in the middle on both cost and complexity. We'll use n8n here because the community edition handles unlimited executions. Your workflow operates in five stages: trigger, image processing, description generation, batch formatting, and delivery.

Stage 1: Trigger and input

Create a simple webhook endpoint in n8n that listens for POST requests. Your team uploads a CSV file or sends a webhook payload with product name, supplier description, and an image URL. Alternatively, you can poll a Google Drive or Dropbox folder every hour for new images.

CopyPOST /webhook/product-listing-auto
Content-Type: application/json { "product_name": "Cotton T-Shirt Classic Blue", "supplier_description": "100% organic cotton, breathable fabric, available in sizes XS to XXL", "image_url": "https://supplier.com/images/shirt-001.jpg", "category": "apparel", "target_marketplaces": ["shopify", "amazon", "ebay"]
}

Stage 2: Image transformation

Once the webhook fires, n8n passes the image URL to Pixelcut AI first. Pixelcut excels at background removal and enhancement, which is your bottleneck. Use Pixelcut's API to remove the background and return a clean, transparent PNG. Next, send that cleaned image to AI Boost for upscaling and any additional retouching. If you're listing apparel, you might request a lifestyle shot or ghost mannequin effect; AI Boost's avatar and outfit-try-on features can generate context-aware product images without needing a photographer.

Copy// n8n HTTP request node to Pixelcut
// Remove background and enhance POST https://api.pixelcut.ai/v1/edit
Authorization: Bearer YOUR_PIXELCUT_API_KEY
Content-Type: application/json { "image_url": "https://supplier.com/images/shirt-001.jpg", "operations": [ { "type": "remove_background", "output_format": "png" }, { "type": "enhance", "enhancement_type": "auto" } ]
}

Once Pixelcut returns the processed image, pass the result to AI Boost. AI Boost doesn't expose a traditional REST API as of March 2026, so you'll need to use their integration layer or save the Pixelcut output to a cloud storage bucket (AWS S3, Google Cloud Storage) and trigger AI Boost via their web interface or partner integrations. If AI Boost is unavailable, use the Pixelcut output directly for your listings; it's already production-ready.

Stage 3: Marketplace-specific description generation

This is where Copy.ai saves hours. Rather than calling Copy.ai once with a generic prompt, create three separate requests within the n8n workflow, each tailored to a marketplace's audience and format requirements.

Copy// n8n HTTP request node to Copy.ai
// Amazon-optimised description POST https://api.copy.ai/v1/generate
Authorization: Bearer YOUR_COPYAI_API_KEY
Content-Type: application/json { "template": "ecommerce_product_description", "inputs": { "product_name": "Cotton T-Shirt Classic Blue", "supplier_description": "100% organic cotton, breathable fabric, available in sizes XS to XXL", "tone": "professional", "marketplace": "amazon", "key_features": ["organic cotton", "breathable", "multiple sizes"], "max_length": 500 }
}

Repeat the request for eBay and Shopify, adjusting the marketplace parameter and tone if needed. eBay benefits from urgency language and stock indicators; Shopify descriptions can be longer and more narrative. Copy.ai returns structured JSON with title, bullet points, and full description.

Copy// n8n HTTP request node to Copy.ai
// Shopify-optimised description with longer narrative POST https://api.copy.ai/v1/generate
Authorization: Bearer YOUR_COPYAI_API_KEY
Content-Type: application/json { "template": "ecommerce_product_description", "inputs": { "product_name": "Cotton T-Shirt Classic Blue", "supplier_description": "100% organic cotton, breathable fabric, available in sizes XS to XXL", "tone": "storytelling", "marketplace": "shopify", "key_features": ["organic cotton", "breathable", "multiple sizes", "sustainable"], "max_length": 1200 }
}

Stage 4: Image resizing and format conversion

Use n8n's built-in image processing nodes or a free service like ImageMagick (via a container node) to generate marketplace-specific image dimensions. Amazon requires 1000×1000 pixels minimum; Shopify prefers 2048×2048; eBay works with flexible dimensions but 600×600 is standard. Create three copies of your processed image at different resolutions and save them to a staging folder in S3 or Google Drive. Tag each file with its marketplace and product ID so the next step knows which file goes where.

Stage 5: Aggregation and delivery

At this point, you have multiple image files and multiple description variants. The final step in your workflow aggregates everything into a single structured JSON output file and optionally a CSV for bulk import.

Copy{ "product_id": "shirt-001", "product_name": "Cotton T-Shirt Classic Blue", "images": { "amazon": "s3://your-bucket/shirt-001-amazon-1000x1000.jpg", "shopify": "s3://your-bucket/shirt-001-shopify-2048x2048.jpg", "ebay": "s3://your-bucket/shirt-001-ebay-600x600.jpg" }, "listings": { "amazon": { "title": "100% Organic Cotton Classic Blue T-Shirt for Men and Women", "bullet_points": [ "Premium 100% organic cotton fabric", "Breathable and soft for all-day comfort", "Available in sizes XS to XXL" ], "description": "..." }, "shopify": { "title": "Classic Blue Organic Cotton Tee", "description": "..." }, "ebay": { "title": "NEW 100% Organic Cotton Classic Blue T-Shirt", "description": "...", "quantity": 50 } }, "processing_timestamp": "2026-03-15T14:32:00Z", "status": "ready_for_review"
}

Save this JSON to your Google Drive, Dropbox, or S3 bucket. Set up a notification in n8n to email your team a link to the output file. At this point, a human takes 2 minutes to review the descriptions for accuracy and approve the listings. No copy-pasting. No manual image resizing. No switching between three different admin panels.

The Manual Alternative

If you prefer to maintain tighter control over each step, run the workflow in stages. Process images with Pixelcut, review them, then feed approved images and your notes to Copy.ai in a separate batch. This adds 15 minutes per product but lets your team catch any issues before descriptions are generated. It's sensible if you're selling luxury goods where tone and accuracy matter more than speed. Alternatively, treat the fully automated output as a draft and require human approval before images and copy are saved to your staging folder. Add an approval node in n8n that waits for your team to click a button before proceeding to image upload and listing generation.

Pro Tips

Rate limiting and API quotas

Copy.ai and Pixelcut both enforce rate limits.

Copy.ai allows roughly 100 requests per minute on standard plans; Pixelcut is more generous at 50 simultaneous operations. If you're processing 500 products, stagger your workflow. Add a delay node in n8n between each product batch, or use a queue system to process items sequentially rather than in parallel. This prevents API throttling and keeps your costs predictable.

Error handling and retries

Not every image will process perfectly. Pixelcut might struggle with complex backgrounds; Copy.ai might generate weak titles on occasionally. Build retry logic into your n8n workflow. Set a maximum of three retry attempts with exponential backoff (wait 2 seconds, then 4 seconds, then 8 seconds before trying again). If a Pixelcut operation fails, log it and continue with the original image; don't let one bad image block your entire batch.

Cost savings through batching

Rather than processing one product at a time, collect 20 to 30 items in a CSV and run them through the workflow together. Batch requests reduce per-item API costs and allow you to negotiate volume pricing with Copy.ai and Pixelcut. If you're processing 500 items monthly, you'll save 15 to 25 percent by batching every two weeks instead of processing daily.

Test with a small subset first

Before deploying this workflow across your entire catalogue, test with 10 products. Review the generated descriptions and images carefully. Adjust your Copy.ai prompts, image dimensions, and tone settings based on what you learn. The difference between a good description and a great one often comes down to tweaking the input template, not the AI model itself.

Store successful outputs for future reference

Save every JSON output file your workflow generates. Over time, you'll build a corpus of working descriptions and image specifications. Use these as templates and examples when adjusting your Copy.ai prompts. You'll notice patterns in what resonates on Amazon versus Shopify, which lets you refine your marketplace-specific instructions further.

Cost Breakdown