Undermind

Undermind is an AI co-researcher for the scientific literature that helps researchers find papers they would otherwise miss and build real command of their field over time. The workflow starts when a user describes what they are working on in plain language. Undermind asks follow-up questions to understand exactly what is needed, then searches the literature by reading and evaluating hundreds of papers, following citation trails until it has found everything relevant. Users can then go deeper together, iterating on reports, diving into full texts, extracting details, and refining their understanding. Undermind keeps tabs on areas of interest and notifies users with important updates through alerts. Every statement can be verified by following in-line citations back to the source paper. The platform supports shared workspaces for collaboration, file and paper libraries, and connections to AI agents like Claude and ChatGPT. Founded by two quantum physics PhDs from MIT with decades of experience in deep research, Undermind is backed by Y Combinator and trusted by research teams at institutions such as MIT, the Koch Institute for Integrative Cancer Research, Northeastern University, and companies including Actipulse Neuroscience and Engage Bio. Pricing starts with a free tier for exploratory research, a Pro plan at $16 per month billed annually with 10x higher usage limits and the latest AI models, a Team plan at $15 per person per month, and enterprise agreements with custom terms, security reviews, and dedicated support. The search experience is designed around the way researchers actually work: the system asks follow-up questions to disambiguate intent, then searches broadly across the literature, evaluating papers for relevance rather than simply matching keywords, and follows citation trails forward and backward until coverage is complete. Results are presented with in-line citations, so any statement can be traced back to its source paper, which is critical for academic credibility. The collaboration layer supports shared workspaces where teams can build paper libraries together, and the agent connection feature lets users link Claude, ChatGPT, and other assistants into the research flow. Alerts monitor user-defined areas of interest and surface important new publications automatically. On the security side, Undermind provides encryption in transit and at rest, independent penetration testing, and guarantees that proprietary R&D data is never used to train models or retained by model providers.