In this study we worked on enriching existing product feeds provided by several companies conducting online commerce in Turkey. Product feeds are used to provide a list of company’s products available for online sale, to advertisement and search service suppliers such as Google and Meta, so that these products can be effectively presented as advertisements and search results to users of the services. However, the product feeds required by these advertisements and search suppliers can be quite long and detailed, whereas the companies providing these details omit filling most of them. The minimal required feed information is a title and description of the product to be sold, which is usually accompanied by a visual of the product. We then proceed to extract the missing information on the product feed from the provided title and description fields, as well as through intelligent queries made on the provided product visuals via pre-trained LLM models. We use 'Gemini-1.5-pro-001' for extracting information from the title and description fields, and ' BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation' for extracting information from the visuals of the product. We show that although results differ based on product type, for most products the title and description fields can be successfully processed by Gemini to provide data for empty fields, whereas BLIP allows extracting 'gender', 'material', and 'color' fields very successfully. Still, we also note that many other fields in a product feed are impossible to be filled by this approach, mostly related to physical properties, such as the length or weight of a product. We finally produce a new description for each product using Gemini, based on the existing and newly extracted fields, using templates of existing successful descriptions. Our efforts in this study aim to improve the search results of e-commerce companies and help advertisement service suppliers in better matching customers and products.