How it works
One open question.
A hundred papers. A grounded hypothesis.
Watch the desktop app take a brand-new, unsolved health problem all the way from a plain-English question to a testable hypothesis, reading the literature and writing back to your files.
/longrun cluster of post-viral patients with chronic fatigue + inflammation, no known cause. Find a likely mechanism and a repurposable drug.
A new, open-ended problem
Dr. Arjun has a cluster of patients with lasting fatigue and inflammation weeks after a mild infection, no known cause, no treatment. He opens Nadhi on his laptop and asks, in plain words.
It pulls & downloads 100 papers
Nadhi searches PubMed, arXiv, bioRxiv and OpenAlex, ranks the open-access literature, and downloads the actual PDFs into his project folder, renamed by author and year.
It reads and indexes every one
All 100 papers are parsed and indexed for retrieval, so every later step is grounded in real sources, not a vague summary.
Specialists brainstorm a mechanism
Sub-agents for immunology, virology, pharmacology and data argue it out across fields and converge on one mechanistic story, with the dissent noted.
A grounded, testable hypothesis
Out comes a specific, falsifiable hypothesis and a repurposable-drug candidate, every claim traced to a paper Nadhi actually downloaded.
Written back to your folder
Nadhi writes hypothesis.md, an annotated reading list, and a draft section into the manuscript on disk. Dr. Arjun stays the scientist who decides what to test.