Overview will disregard any charts and formatting - the first step of the process is to extract only the text from the documents (by uploading PDFs to documentcloud, for example).
If there are patterns in the text, the clustering algorithm should group documents with similar patterns in the same folders. Do you have a set of documents in mind that you can try?
_jonas