After working on biomedical literature search, discovery and recommender web applications for many months I concluded that extracting, structuring or synthesizing "insights" from academic publications (papers) has negligible value in industry.