ProtocolNerd is a large language model-based assistant that helps scientists find the right experimental protocol. A researcher describes an experiment in plain English, and the system finds and ranks the most relevant protocols and papers, explaining why each result fits and what it may not cover. It currently supports biology and chemistry.
ProtocolNerd turns a request into a structured experiment profile, asking a targeted clarifying question when an essential detail is missing. It then searches a curated corpus of Protocols.io protocols with both keyword and semantic search, alongside the scientific literature: PubMed for biology and Europe PMC for chemistry. Protocols and papers compete in one ranking ordered by how well each matches the experiment, and every result is annotated with why it matches, what may not fit, and which details remain to be checked.
Researcher
NYU Courant Institute School of Mathematics, Computing, and Data Science
Professor of Computer Science
Courant Institute of Mathematical Sciences, NYU
