The fourth wave means that, for the first time, machines can reliably classify documents and extract data, without the need to previously train with document samples