Agent was just doing eDiscovery processing and not searching.
Processing has to be completed before document review (searching, etc).
Processing: Before you can search documents you have to extract their content: text, metadata, and attachments (attachments and embedded items get recursively processed to). Document formats are identified while processing too, using binary signatures, like: OutlookEmail, EML, Word2003, Excel2007, etc. because people try to hide shit by giving documents false extensions (ex: changing Word .doc extension to .jpg extension), thinking the software won't recognize file type.
The indexing of text comes after processing (or merging of indexes if distributed processing was done).
Then agent could do full-text search of all documents.
Sounds like this agent wasn't searching, he was just looking at processing dashboard screen (or log) that is giving updates about file types found, number of total documents, etc.