We don't have enough scholars to analyse Puratan Birs
The idea
An AI research assistant for the few scholars who can analyse puratan birs. Its first job is the foundation for everything else: turning each manuscript page into reliable, machine-readable text, with scholars verifying every line. Alongside that, it highlights what's distinctive about the hand and the page, and lines a bir up against others, showing its working at every step so the scholar decides. The challenge goes well beyond off-the-shelf text recognition: reading old handwritten Gurmukhi with no spaces between words, showing exactly where the AI is unsure, and making verification fast enough to cover whole birs.