Mimir Systems

Mimir Systems is an AI-powered Q&A tool built for researchers in materials science, chemistry, polymers, physics, and related physical sciences. Users type a technical research question in plain language and receive an answer with citations drawn from over one million indexed scientific papers. The core technical distinction is architectural. Mimir uses retrieval rather than generation for its citations. When a user asks a question, the system searches its index of published papers, figures, tables, and numerical data, then assembles an answer from what it finds. If a paper does not exist in the index, the system cannot cite it. This eliminates the citation hallucination problem that affects general purpose language models, where tools like ChatGPT sometimes produce references to papers, authors, or DOIs that do not exist. The indexed corpus currently covers materials science and engineering, carbon fibers, polymers, composites, ceramics, glassy materials, aerospace materials, chemistry, and physics. Coverage is deepest in materials science, with expansion into adjacent fields ongoing. The tool is free for users with .edu and .gov email addresses during the current beta period. An enterprise product is also available for R&D teams at companies and institutions that want to index proprietary internal documents (patents, lab reports, technical memos) alongside the public literature.