Help Center & Documentation

An idea check typically takes 15 to 40 minutes, depending on the volume of published literature. A paper review takes 20 to 45 minutes for full PDF parsing and two independent reviewer audits. Reference checks run in under 2 minutes. You can safely close your browser; runs continue asynchronously in the background and notify you when complete.

Every factual assertion in a NoveltyAxis report is strictly categorized and verified word-for-word against retrieved full-text documents. We use three explicit badges: Evidence (exact quote verified in published text), Inference (transparent reasoning step derived directly from evidence), and Hypothesis (untested proposed direction). We never allow ungrounded claims.

No automated tool can provide absolute mathematical proof of novelty. NoveltyAxis maps the exact frontier of published work, identifies the closest competing baselines, and outlines the precise delta between prior work and your proposition so you and your peer reviewers can evaluate novelty with total clarity.

NoveltyAxis searches four open scholarly indexes: arXiv, OpenAlex, Semantic Scholar and Crossref. When a paper is open access (an arXiv preprint or an open-access PDF), it reads the full text; otherwise it works from the abstract and metadata, and says so.

Only you. Your research questions, draft manuscripts, and generated reports are strictly private to your account and encrypted in transit and at rest. We never use your drafts or ideas to train language models, and you can export or permanently delete all your data at any time under Account settings.

All literature findings, paper quotes, and audit notes retrieved up to that point are saved immediately to disk. You can open any paused run and click Continue to pick up exactly where it stopped without consuming an additional monthly run allowance.

Each plan provides a monthly quota of comprehensive runs. One run equals one complete Idea Check or one comprehensive Paper Review. Reference checks are free on every plan, up to 50 a month. You can upgrade, downgrade, or cancel at any time, and you always retain full access to all your past reports.

Yes. Hallucinated citations created by generic AI tools typically fail to resolve in Crossref or OpenAlex, or have mismatched authors and years. Our Reference Checker queries authoritative databases directly without a language model, flagging likely fabricated citations and broken DOIs in about two minutes.

Submissions are independently evaluated by two separate reviewer agents on a five-point rubric: novelty, quality, relevance, verifiability and clarity, each with a written justification. Every critique quotes the exact line in your manuscript with suggested revisions.