Initially randeli is being developed as a python-based CLI. But a web-based UI is on the mid-term roadmap.
Earlier versions were developed in Go and C++, but those have been postponed to get the first release out.
As are usability features to help users find the best "augmentation" styles for them, automatically.
I think there is also scope to add alternative augmentation (audible?) as well.
Currently randeli does not work with PDFs generated by XeLaTeX,
these use a different font technology in the PDF (CID vs Type1)
that makes it harder to process the text (glyphs are not a simple
index into UTF-8 charset). This is the highest priority. A workaround
has been commited that uses OCR to augment the PDF bypassing the use
of text/font details.