This one is answered in one of the reddit post.
The OCR that will be released in the beta will not work across the whole database (note search) but this is something we want to test at some point. OCR operations are onerous in terms of CPU usage so we have to explore when scanning attachments and use some level of caching to have a performing search.