Openstore can via AI summarize a text and use this to give users suggestions for images that might fit the text they are writing.
Pain: Ever since the merger the pressure to meet tight deadlines has only increased and efficiently sourcing images remains a bottleneck. Our editors collectively spend hours each day scrolling through vast image libraries and searching external stock platforms, often settling for visuals which are only adequate. This not only drains time but also risks diluting the quality and impact of the final article.
Pill: Openstore, Openstore API, Openstore AI Search
Description: Iceman Media’s Openstore AI Search transforms traditional image search by understanding the individual meaning of each article and presenting only images which are highly relevant to that article’s specific content, tone and perspective. This means that instead of relying on manual image tags and keywords, our solution identifies the deeper context and intent not only of the individual content but also of all available imagery for the piece. These are matched and evaluated by powerful ranking engines, resulting in a discrete and highly curated selection of images, presenting editors with a set of tailored options that feel like they were chosen by an expert.
Proof: Integrating AI search into our editorial workflows via Openstore was a seamless add-on and has been incredibly effective in reducing the manual and mental workload of finding relevant and appropriate images for publication. This has not only reduced the amount of time we spend searching for images but also the amount of time we spend downloading and adjusting images from external agencies. This is because we often have exceptional images in-house, particularly historical, which were unavailable due to limited or missing image data. Overall, this project has worked very well. Our workflow efficiencies and the reduced use of external agencies have consistently combined to help us reduce our costs while maintaining quality and improving publication speed.
User edits article/text in CMS system
CMS system sends article text to OpenstoreAPI
Openstore uses AI to obtain an *embedding vector for the article
The embedding as vector is then used to get images with similar embeddings vector via Openstore image search
The suggestions of images is presented to the user and user place the image of choice to the article
*) Text embedding (the same as word embeddings) is a transformative technique in natural language processing (NLP) that has improved how machines understand and process human language. Text embedding converts raw text into numerical vectors, allowing computers to understand it better.
