2 comments

[ 3.7 ms ] story [ 5334 ms ] thread
Nice project. Ideally it should be possible to run arbitrary graph or datalog queries to get relevant items but a hierarchical organization w/ basic content type tagging + vector similarity search is a good starting point.