WorkBuddy E-commerce Asset Classification: Did It Mistake Filenames for Product Images?

I came across a roundup of e-commerce AI image generation tools for 2026 claiming they double efficiency. Personally, I feel that the faster images are generated, the messier single-person stores become. Recently, I've been using WorkBuddy to organize product assets for Shopify and Xianyu (Idle Fish). I want to use it as a "packer" before listing: dumping main images generated by Gemini Nano Banana and Flux Art into folders, letting it categorize them by SKU, and then organizing colors, sizes, and prices into a spreadsheet.

Let me do the math on savings: if this step works smoothly, I save at least 40 minutes a day, roughly 20 hours a month.

But after running it for two days, I'm confused. The first time, I uploaded over a dozen images and asked it to identify the products and fill out the form. It took the filename "Red Style 01" literally as the color, even though the image clearly showed hazy blue. The second time, I changed all filenames to codes and told it to look only at the images. It did generate a table, but the price column was full of "Unknown," and it threw an error saying "Unable to parse image metadata." I also tried having it OCR text first and then merging tables, but complex layouts still resulted in errors; gift descriptions often ended up in the notes field.

I'm not sure if I misconfigured something or if WorkBuddy just isn't suited for interpreting image content. It handles Excel quotes and customer service logs obediently, but gets skewed when dealing with images. Has anyone used it for bulk archiving product images? Or does anyone know if I should integrate an OCR alternative first?