Researchers utilizing a proprietary deep-learning model have successfully deciphered fragmentary copper plates found in Karnataka, written in the 9th-century Nandinagari script. The texts, analyzed in a report for Heritage Science, contain detailed protocols for managing subterranean water tables and preventing soil salinization in arid regions. The manuscripts describe a system of interconnected step-wells and clay filtration pipes designed to recharge aquifers during monsoon surges.
This discovery provides the first written evidence of the engineering logic behind the complex hydraulic systems found at medieval South Indian sites. The AI model was able to bridge gaps in the eroded inscriptions by comparing the grammatical structures to contemporary Shilpa Shastra texts, revealing a sophisticated understanding of hydro-geology and the chemical properties of different soil layers for water purification.