Ask Mendeleev about materials —
literature you've missed and candidates you haven't tried.
Search the Indian materials-science literature, find what's been measured, propose computationally stable candidate materials, and predict their properties with a universal ML potential — in a single conversation.
Mendeleev is a research copilot built for materials scientists with a deliberate focus on the Indian materials research context. It connects published literature from Indian institutions, the structured chemistry inside each paper, and a ~300,000-entry computed candidate-materials database. An LLM with ten tools picks the right operation for your question.
Why India. Indian materials research is large and active but distributed across hundreds of institutions; the gray literature and Indian-domestic sources are poorly indexed globally. Mendeleev starts from this corpus first, with plans to add Shodhganga theses, CSIR/DRDO open output, and Indian-domestic journals through partnerships.
What's missing (honest). v0 corpus is recent (2020+) and weighted toward published papers. PhD theses, internal reports, and pre-2020 work aren't indexed yet. ML predictions are MACE-MP-0 only — energy and forces; bandgap and elastic properties need additional models. There's no user accounts, no saved workspaces, no collaboration features. Yet.
Built on open infrastructure: OpenAlex (corpus), LeMaterial / Materials Project / AFLOW / OQMD (candidates), MACE-MP-0 (ML potential), pymatgen (chemistry), Gemini 3.1 Flash-Lite (LLM), BAAI/bge-small (embeddings), Let's Encrypt (TLS). Hosted on AWS Lightsail Mumbai. Free for Indian academic use; contact for institutional or industrial access.
Contact: api@natarajalabs.com