askchem-7d0107d9·1 events·first seen Aliases: AskChem
AskChem is a new retrieval infrastructure that reframes chemistry literature search around atomic, provenance-carrying claims rather than document rankings, indexing 2.4M claims from 147K papers. Each claim is grounded with a source DOI and verbatim quote, and the system exposes a faceted taxonomy, evidence graph, and living taxonomy for synthesis. The system provides REST, SDK, and MCP access for AI agents, and benchmarking shows that grounding GPT-5.5 in AskChem yields 100% resolvable DOIs versus 88.3% without retrieval. The work is relevant both as a domain-specific RAG infrastructure paper and as an example of MCP-native scientific tooling for AI agents.