Bhashita Wants to Teach AI How Sanskrit Actually Works

Field notes

The proposed pipeline moves from damaged palm-leaf or paper input through restoration, script recognition, transcription, grammatical parsing, commentary linking, and search. Its linguistic layer treats Pāṇinian rules as a validation constraint, while the retrieval layer is intended to connect root texts with bhāṣya, ṭīkā, and tippani traditions. The page names five historical scripts and shows an Ayurveda case built around a Gudūcyādi Varga verse, giving the broad manuscript claim a specific test subject.