Transform product images into marketplace listings with descriptions and SEO metadata

Imagine launching 200 new products across Amazon, eBay, Shopify and your own website. Each one needs a unique description written in platform dialect, optimised images resized to spec, and SEO metadata tailored to category rules. Now imagine your team spending 15 minutes per product doing this manually. That's 50 hours of work for one catalogue push, and you've already lost the first-mover advantage. Product image workflows are one of the slowest bottlenecks in e-commerce. The work feels simple until you're doing it at scale: crop the image, write the headline, compose three variations of the description, generate alt text, add keywords, upload to each channel. Repeat 200 times. Your margins shrink because labour costs spike, and your listings stay generic because nobody had time to optimise them properly. This workflow automates the entire chain. You upload a product image and basic details once. Within minutes, you get marketplace-ready images, platform-specific descriptions, and SEO metadata for every channel. No manual handoffs. No copy-pasting between tools. For more on this, see E-commerce product catalogue optimization and listing exp....

The Automated Workflow

You'll need an orchestration platform to connect the pieces. We'll use Make here because it handles file uploads smoothly and integrates cleanly with both image and text AI services, but Zapier works fine too if you prefer it. The flow moves like this: trigger from your e-commerce backend or a simple form, send the product image to Pixelcut AI for background removal and optimisation, pass the cleaned image to AI Boost for upscaling and marketplace-specific resizing, feed product details and the image analysis to Copy.ai to generate descriptions and metadata, then post everything to your chosen platform via API.

Setting up the trigger

Start with a webhook or form submission. Most e-commerce platforms let you trigger workflows when a product is created or flagged for listing. If you're using Shopify, you can listen for product creation events; if you're on a custom system, set up a simple Make webhook that accepts POST requests.

CopyPOST https://hook.make.com/your-webhook-id
Content-Type: application/json { "product_id": "SKU-12345", "product_name": "Wireless Headphones Pro", "category": "Electronics", "base_price": 89.99, "image_url": "https://your-storage.s3.amazonaws.com/raw-images/product-001.jpg", "marketplace": "amazon"
}

Make receives this and starts the workflow. Store the product ID and name in variables; you'll need them later to keep everything linked.

Image processing stage one: Pixelcut AI

Pixelcut removes backgrounds and enhances the image. Most product photos have cluttered backgrounds; removing them increases marketplace visibility and lets you add consistent branded backgrounds later. Configure a Make module to call Pixelcut's API:

CopyPOST https://api.pixelcut.app/v1/edit/background-remove
Authorization: Bearer YOUR_PIXELCUT_API_KEY
Content-Type: application/json { "image_url": "{{trigger.image_url}}", "format": "png", "quality": "high"
}

Pixelcut returns the cleaned image URL. Store this in a Make variable called cleaned_image_url. You now have a background-free product image ready for the next stage.

Image processing stage two: AI Boost

AI Boost handles upscaling and marketplace-specific resizing. Amazon wants 1000x1000px minimum; eBay prefers 500x500px. Rather than manually resizing for each platform, send the cleaned image to AI Boost and request multiple output versions.

CopyPOST https://api.aiboost.io/v1/upscale-and-resize
Authorization: Bearer YOUR_AIBOOST_API_KEY
Content-Type: application/json { "image_url": "{{cleaned_image_url}}", "upscale_factor": 2, "resize_outputs": [ { "name": "amazon", "width": 1000, "height": 1000 }, { "name": "ebay", "width": 500, "height": 500 }, { "name": "shopify", "width": 600, "height": 600 } ]
}

AI Boost returns URLs for each resized version. Create a Make array variable to store all three:

Copy{ "amazon_image": "https://api.aiboost.io/output/amazon-001.jpg", "ebay_image": "https://api.aiboost.io/output/ebay-001.jpg", "shopify_image": "https://api.aiboost.io/output/shopify-001.jpg"
}

At this point, your images are optimised and ready. Next comes the text.

Content generation with Copy.ai

Now you have clean, upscaled images. Pass product details and the image to Copy.ai to generate descriptions. Copy.ai can write in different styles and tones; you configure it once and it adapts to each marketplace automatically. Create two separate Copy.ai requests: one for marketplace listings, one for SEO metadata. For the main description:

CopyPOST https://api.copy.ai/v1/generate
Authorization: Bearer YOUR_COPYAI_API_KEY
Content-Type: application/json { "template": "ecommerce_product_description", "inputs": { "product_name": "{{trigger.product_name}}", "category": "{{trigger.category}}", "price": "{{trigger.base_price}}", "marketplace": "{{trigger.marketplace}}", "tone": "professional_persuasive", "max_length": 500, "include_image_analysis": true, "image_url": "{{cleaned_image_url}}" }
}

Copy.ai analyses the image and generates a marketplace-appropriate description. Store the output:

Copy{ "description": "{{copyai_response.text}}", "bullet_points": "{{copyai_response.bullets}}"
}

For SEO metadata, run a second Copy.ai request:

CopyPOST https://api.copy.ai/v1/generate
Authorization: Bearer YOUR_COPYAI_API_KEY
Content-Type: application/json { "template": "ecommerce_seo_metadata", "inputs": { "product_name": "{{trigger.product_name}}", "category": "{{trigger.category}}", "description": "{{first_copyai_response.text}}", "include_keywords": true, "keyword_count": 8 }
}

This returns SEO-optimised title, meta description, and keywords. Your Make workflow now has all the data needed to post listings.

Final step: post to marketplace APIs

Use Make's HTTP module to send formatted data to your marketplace platform. Here's an example for Amazon:

CopyPOST https://mws.amazonservices.com/Products/2011-10-01
Authorization: AWS4-HMAC-SHA256 ...
Content-Type: application/xml <AmazonEnvelope> <Header> <DocumentType>Product</DocumentType> </Header> <MessageType>Product</MessageType> <Message> <MessageID>1</MessageID> <OperationType>Update</OperationType> <Product> <SKU>{{trigger.product_id}}</SKU> <StandardProductID> <Type>ASIN</Type> <Value>{{trigger.asin}}</Value> </StandardProductID> <ProductData> <Home> <Title>{{seo_response.title}}</Title> <Brand>YourBrand</Brand> <Description>{{copyai_description}}</Description> <BulletPoint>{{copyai_bullets[0]}}</BulletPoint> <MainImageURL>{{amazon_image}}</MainImageURL> <SearchTerms>{{seo_response.keywords}}</SearchTerms> </Home> </ProductData> </Product> </Message>
</AmazonEnvelope>

Each marketplace has its own endpoint format. Create separate posting modules for Amazon, eBay, and Shopify. Use Make's conditional logic to route data based on the marketplace field from the trigger. After posting, log the success response and store the marketplace listing ID back to your database. This way, if you need to update or delete the listing later, you have the reference.

The Manual Alternative

You can run most of this without orchestration if you prefer tighter control. Download images from Pixelcut and AI Boost manually, paste product details into Copy.ai, review and edit the output, then upload to each marketplace yourself. This gives you time to refine descriptions and catch errors, but you'll spend 10-15 minutes per product instead of the workflow's 2-3 minutes. Choose this route if you have a small catalogue (under 50 products) or if you want human approval on every listing before it goes live.

Pro Tips

Rate limiting matters.

Copy.ai and AI Boost both enforce request limits on their free tiers.

If you're launching 200 products at once, you'll hit limits fast. Stagger requests in Make by adding a 2-second delay between Copy.ai calls. Use paid plans if you're processing more than 50 products weekly.

Always validate image URLs before posting.

Pixelcut and AI Boost sometimes return invalid URLs if the source image is corrupt or too small. Add a Make HTTP GET module to check each URL before posting. If it fails, send yourself a Slack notification so you can investigate.

Cache descriptions where possible.

If you have multiple products in the same category, Copy.ai might generate very similar descriptions. Run your Copy.ai output through Claude Opus 4.6 to add unique selling points or reword sections. Claude is cheaper for light text tweaks than running Copy.ai multiple times.

Test with one product first.

Don't launch 200 products on day one. Run the workflow on three test products, review the listings manually, then adjust your Copy.ai prompts. A small tweak to tone or style can massively improve real-world performance.

Monitor costs per batch.

Each image processing call costs money. Track how much you spend per product batch so you know if the ROI makes sense for your margin. If you're selling £5 items, spending £0.50 per listing on automation eats profit. For £50+ items, automation is almost always worth it. For more on this, see Real estate listing automation from property inspection r....

Cost Breakdown