TreasureX Researchcs.AIAskChem: A New Way to Search Chemistry Papers by Claims, Not Just Documents
Bing Yan, Gregory Wolfe, Stefano Martiniani, Kyunghyun Cho · July 30, 2026
Searching through chemistry literature is like trying to find a needle in a haystack—except the haystack is millions of papers, and the needle is a specific fact buried in one of them. Traditional search engines return lists of papers, forcing scientists to read through each one to find the relevant bits. AskChem, a new tool from researchers at NYU and Meta, flips this model: instead of returning papers, it returns individual claims—atomic, fact-like statements—each traced back to its source paper and exact quote. This means you can search for a specific chemical property or reaction and get a direct answer with proof, not just a list of possibly relevant documents. The system currently indexes 2.4 million claims from 147,000 papers, and it's designed to work with AI agents via APIs, making it a powerful tool for automated literature review. In tests, grounding an AI reader with AskChem improved the accuracy of finding source DOIs from 88.3% to 100%, and it produced more citations than other systems. For investors, this represents a step toward making scientific knowledge more accessible and verifiable, potentially accelerating R&D in chemistry and materials science. For everyday readers, it's a glimpse into how AI can transform how we interact with scientific information. The tool is live at askchem.org. This summary is based on the arXiv paper: https://arxiv.org/abs/2607.28618.
Educational content only, generated with AI assistance from public research abstracts. Not investment advice. Thank you to arXiv for use of its open access interoperability.