Efficiently organizing product images is essential for enhancing their discoverability and improving user experience. However, manual image categorization is often tedious and inconsistent, making automation a game-changer for businesses dealing with large volumes of media assets.
In this blog post, we’ll discuss how Cloudinary’s AI Vision Add-on and MediaFlows provide a seamless solution for automatically categorizing images based on prompts, ensuring your media library stays organized and searchable as it scales.
E-commerce platforms host diverse product images, each serving a unique purpose. For instance:
- Lifestyle. Showcases products in the real world.
- Model. Displays products worn or used.
- Selection Chart. Provides comparison charts or guides.
- Off-Figure. Captures standalone product images.
The Cloudinary AI Vision Add-on leverages advanced AI to analyze images based on custom-defined prompts. It can automatically tag product images with relevant categories, enabling businesses to maintain a structured asset repository without extensive manual work.
When an image is uploaded, AI Vision analyzes it based on predefined prompts and applies the relevant tags. Below is an example API request to classify an e-commerce product image:
POST https://<API_KEY>:<API_SECRET>@api.cloudinary.com/v2/analysis/{cloud_name}/analyze/ai_vision_tagging
{
"source": {
"uri": "https://res.cloudinary.com/demo/image/upload/v1741748003/engin-akyurt-ehdI_89nzMo-unsplash_mqnnoh.jpg"
},
"tag_definitions": [
{ "name": "lifestyle", "description": "Does the image feature a real-world scene?" },
{ "name": "model", "description": "Is a model present in the image?" },
{ "name": "selection-chart", "description": "Does the image contain a comparison or selection chart?" },
{ "name": "off-figure", "description": "Is this a standalone product image?" }
]
}
Code language: JavaScript (javascript)
After processing, the response will return a list of applicable tags, ensuring accurate classification.
{
"tags": { "name": "model" }
}
Code language: JSON / JSON with Comments (json)
To streamline the categorization process, Cloudinary’s MediaFlows automates tag updates on asset upload. MediaFlows provides a visual, low-code or no-code interface to integrate AI Vision into workflows.
After logging in to Cloudinary, select MediaFlows from the multi-product navigation bar on the left. Then, go to Flows > New PowerFlow and follow these steps:
- Trigger. Use the Cloudinary Upload block to detect when an image is uploaded. Once this block is added, ensure the Webhook is set.


- AI Vision. Use the AI Vision Tag By Prompts block to configure the prompts and tags.
- Test. Upload an asset to the DAM.The live logs within MediaFlows should show this event:

The tag for the uploaded image should be updated with the tags detected by AI Vision.

Beyond simple image categorization, MediaFlows enables businesses to build more complex workflows that integrate with AI Vision and other AI capabilities.
Examples:
- If the asset was uploaded to the “Product” folder, then AI Vision needs to categorize images into: Lifestyle, Model, Selection-Chart, and Off-Figure.
- Set this as a Structured Metadata instead of a tag.
- If the value detected is “Off-Figure”, then remove the background.
- If the asset was uploaded to the “Content” folder, then AI Vision needs to categorize images into: Logos, Guidelines, Event Photography, Storefront, and Merchandise.
- Set this as tags.
- If “Event Photography”, send this to the manual moderation queue.

By leveraging these capabilities, businesses can design intelligent media workflows that go beyond simple tagging and enable smarter automation across the entire content pipeline.
By leveraging Cloudinary AI Vision Add-on and MediaFlows, businesses can automate image categorization with high accuracy and minimal manual effort. This AI-driven approach ensures a well-organized media library, improving searchability and enhancing content workflows.
Ready to get started? Explore Cloudinary’s AI Vision Add-on and MediaFlows to transform your e-commerce image management today!