About litscout

What it is

Most literature search is keyword search with citation counts bolted on. You type words, you get documents containing those words, and relevance is decided by how often other people cited them. That works when you already know the vocabulary of the field. It works badly when you are entering one, when the idea you are chasing is described three different ways by three communities, or when the paper you need is a 2019 preprint nobody has cited yet.

litscout starts from intent instead. It reads what you are actually trying to find out — including messy input, half-remembered titles, a pasted abstract, a DOI, or a description of the problem rather than the topic — and builds a structured picture of the objective before it searches anything. Retrieval then runs along several paths at once: exact terms, meaning, and the citation graph around whatever you already know is relevant.

The ranking is the product

Finding candidates is the easy half. The engine's job is deciding which of them answer your question, and being able to say why. Every result carries an explanation of what matched — the requirement it satisfies, how it relates to a paper you already flagged, where it sits in the citation neighbourhood.

That constraint shapes the whole system. Retrieval and ranking stay deterministic and explainable wherever they can be. Language models are used for two narrow jobs — pulling references and requirements out of what you wrote — and not for deciding what is relevant, because a ranking you cannot interrogate is a ranking you cannot trust for research.

What it searches

Journal papers, conference papers and preprints, plus the artifacts that usually fall outside a paper search and often hold the answer:

• Patents, standards and technical reports
• Theses and dissertations
• Datasets and benchmark suites
• Books and book chapters

Reports

Beyond search, litscout writes research reports in three depths — a short summary, a section-by-section detailed report, and deep research that explores recursively — in any of six tones, from objective through to critical. Reports cite their sources and the sources are yours to check.

What it deliberately does not do

It is not a chatbot, and it is not an autonomous research agent that decides things for you. It does not read papers on your behalf and tell you what they conclude, it does not judge methodological quality, and it does not check whether a paper has been retracted or superseded. A ranked list is not an evidence review and a relevance score is not a finding — which is why the Usage Policy prohibits using it as the sole basis for decisions about a person's health, legal position, finances, employment, housing, insurance or education.

Your queries, notes and uploads are never used to train any model. That is a commitment in the Privacy Policy, section 7, not a marketing line.

Who runs it

litscout is operated by Shagufta Shaheen, based in Mango, Jamshedpur, Jharkhand 831012, India. It is a small, independent operation rather than a venture-funded team, which is the reason for two things you may notice: the pricing is low, and the published documents are unusually specific about what the service does and does not guarantee.

The full operator identity and postal address are in the Legal Notice. For anything else, write to us — a person reads it.