Stage 9 · Write-Up

From results to manuscript, without losing the receipts.

AI-drafted sections grounded in your actual project data, a citation library with five switchable styles, reference verification against Crossref, and export to Word, Markdown, or LaTeX — including your own journal template.

Drafting that starts from your data, not a blank page

The write-up assistant drafts standard manuscript sections — introduction, methods, results, discussion — from what actually happened in your project: the design you built, the trials you ran, the coding schemes you applied, and the analyses you computed. Sections are generated one at a time, saved server-side as they are produced, and survive reloads and navigation. Each section carries its own regenerate and delete controls, so the draft evolves under your direction.

Because the methods section is generated from recorded facts rather than memory, it states what a reviewer needs: exact model versions, parameters, replication counts, prompt strategy, coding procedure, and reliability statistics.

Citations treated as first-class data

  • Curated citation library

    A built-in library of LLM-research methodology papers, plus your own custom citations per project. In-app methodological guidance cites its sources with the same references.

  • Editable BibTeX

    The references section is a real BibTeX file, assembled on finalize from citations detected in the prose plus your manual entries. Regenerating merges by key — it never clobbers your hand edits.

  • Five citation styles

    APA, MLA, Harvard, Chicago author–date, and numbered Vancouver, switchable per project and rendered server-side, reflected in every export format.

  • Reference verification

    Every reference can be checked against Crossref and the open web. AI-supplied citations that cannot be confirmed are explicitly marked unverified — hallucinated references don't slip through.

  • Find literature

    AI-assisted literature search drafts candidate related work with real, verifiable references to seed your bibliography.

  • Exhibits from analysis

    Charts from your analyses — including ones produced in AI Analyst conversations — can be attached to the manuscript as numbered figures with captions.

From draft to submission

  1. Generate and edit sections

    Draft each section with AI, edit freely, and regenerate selectively. The draft is persisted per project and only replaced when you say so.

  2. Finalize references

    On finalize, AIScholar assembles the BibTeX from prose-detected and custom citations, runs verification, and renders the bibliography in your chosen style.

  3. Export in your target format

    Word and Markdown export for collaboration, or LaTeX with figures and bibliography bundled — compiled server-side, with support for uploading your own journal's .tex template with placeholder substitution.

The draft is a starting point — by design

AIScholar drafts prose; it does not submit papers. The scientific claims, framing, and final wording remain yours, and the methodological warnings that accompanied your study travel with the draft so known limitations end up in the limitations section, not under the rug.

Everything the manuscript claims is backed by the audit trail described in reproducibility.

Turn a question about LLMs into published research.

Start your first study today. No human participants required.