Undermind, founded in 2024 by two MIT-trained quantum-physics PhDs and backed by Y Combinator, takes a conversational, iterative approach to literature search: a user describes their research in plain language, and the system asks follow-up questions to refine understanding before evaluating hundreds of papers and following citation trails to build a curated result set.
That follow-up-question refinement, plus full-text search that goes beyond only scanning abstracts, aims directly at a common failure of keyword search: a vague or overly broad initial query returning results that miss what the researcher actually meant. Undermind also states explicitly that user data is never used to train its underlying models, a specific commitment aimed at researchers wary of proprietary or unpublished work being absorbed into a training set.
A free tier covers standard use with rate limits, and paid plans start around $16 a month for substantially higher usage limits. For a researcher whose initial query is hard to express precisely in a single search string, Undermind's conversational refinement process addresses that ambiguity more directly than a one-shot keyword search would.
