What do you like best?
ReadCube has a slick interface and its abilities of reading, highlighting and annotating PDFs are very useful. When importing PDFs, ReadCube tries to identify the relevant metadata, similarly to Mendeley. Unfortunately, many academic publishers use different formats for incorporating metadata in their articles, making automated solutions like ReadCube fail often. However, in addition to allowing manually entering the metadata, ReadCube adds a welcome functionality, called "Help ReadCube resolve article", where the user fills the blanks by simply highlighting parts of the article and setting its category (title, authors, journal, etc.). Then ReadCube checks online for a match. This makes correcting PDF metadata in a large libraries much easier.
What do you dislike?
In comparison with its competitors (EndNote, Mendeley, Zotero, etc.), ReadCube is still a young system and there is a lack of support for the many different reference styles in use. Also, the basic (free) version has limited capacity for adding multiple PDFs, requiring manually entering one by one.
Recommendations to others considering the product
ReadCube can be great tool for centering your reading and annotating activities on PDF files, but regarding reference management, it still falls a bit short, probably due to its relatively young age. Nonetheless, it is a platform that deserves following, since in its short history, it introduced some interesting innovations on its area.
What business problems are you solving with the product? What benefits have you realized?
The main benefits of ReadCube is its interface, more modern and clear than its predecessors, and its transparent access to internet resources, which makes easier to find matches for orphan PDFs.