Nephilim

This project introduces an AI-native drug discovery workspace that connects traditional medicinal knowledge with modern computational biology. Many medicinal plants documented in Ayurvedic and ethnobotanical traditions remain molecularly understudied, with fragmented data on their bioactive ligands and protein targets. Our platform integrates a structured database of plants, associated ligands, and curated traditional knowledge, forming a searchable foundation for hypothesis generation. On top of this, Through structured context engineering, the agent maintains coherent reasoning across extended sessions. The workspace is fully interactive and human-collaborative, featuring molecular visualization, literature integration, hypothesis sketching, code inspection, and file management within each session. Results are compute-cached per input combination to ensure scalability and reuse.