Getting started: AI thesis writing, grounded
The 30-minute path through CiteDash AI's thesis writing tool: pin a research question, build a small library, and draft a section where every claim cites a real paper.
CiteDash AI is an AI thesis writing tool organised as sixteen connected tools that follow the whole thesis workflow, but you don't need all sixteen on day one. This guide walks the shortest useful path: a project, a pinned question, a handful of papers, and one AI-drafted section you can actually defend.
1. Create a project and set your deadline
From the home screen, create a project and pick your method (literature-based, theoretical, or empirical). The Roadmap tool builds the milestone plan for that method immediately, and if you set a deadline every milestone gets a target date, so from the first minute you can see whether you're on pace.
2. Pin a research question
Open Question Builder and describe your interest in plain language: "something about spaced repetition and exam performance" is enough. You'll get back candidate questions with evidence, novelty, and feasibility signals drawn from real literature. Already have a question? Type it in and pin it directly.
The pinned question matters more than it looks: every drafting tool frames its output around it, so pin before you write.
3. Build a small library
Open Literature Finder and search your topic. Results come from CiteDash's own index plus a live sweep of OpenAlex, PubMed, Semantic Scholar, and arXiv, reranked into one list. Save the papers that matter. Open-access PDFs are fetched, parsed, and indexed automatically. For paywalled papers you have legal access to, upload the PDF yourself in the Thesis Library.
Aim for five to ten papers with full text before drafting. Full text is the raw material for everything downstream: only claims grounded in a stored PDF can be verified.
4. Draft one section, and read the colors
Open Thesis Editor, create a chapter, and use Write with AI on a single section. The draft is generated from your library's full text only, and each claim arrives with a citation that resolves to a real paper. The Fact Checker then verifies each cited sentence against its source before you see it.
Green means the source supports the claim; flagged sentences mean the AI could not ground that statement in your library. Rewrite them, add a source, or delete them. That honesty is the product working, not failing.
Where to go next
- Import your existing reference library (see the BibTeX import guide).
- Build an evidence matrix in Synthesis Lab once your library passes ~15 papers.
- Check Submission Readiness any time for an honest picture of what remains.