AI-Powered Tagging

The problem

Product images are often missing alt text, which leaves screen reader users without a description, and missing metadata, which makes images hard to find and manage. Writing both by hand for large catalogs takes a lot of time.

My role

I’m developing the concept and designing the workflow.

The approach

SmartTag AI uses AI to analyze product images and produce two kinds of tags at once: descriptive alt text for screen readers, and metadata such as product name, category, and features for SEO and content management.

The workflow:

1. Upload. The user selects product images, which are checked for resolution and a clear subject.
2. Image recognition. A pre-trained model identifies the product and its features, such as shape, color, and setting.
3. Alt text. Natural language processing turns what the model sees into concise, meaningful alt text.
4. Metadata. The system generates category, name, and feature tags.
5. Review and edit. A person reviews every alt tag and piece of metadata and refines them for accuracy.
6. Apply. Approved tags are written to the image metadata or CMS and the product listings are updated.

The result

The project is still in the testing and learning phase. Its goal is a dual-tagging tool that improves the experience for people with disabilities while making product content easier to search and organize.