Write like a scholar.
Edit like a reviewer.
Hanova AI is a step-by-step research writing studio for PhD and research scholars. Define your topic, find high-impact papers, extract their methods and findings, design your methodology, then outline, write, humanise and cite your paper — reviewing and refining every step yourself.
-
1TopicResearch questions and search queries
-
2LiteratureHigh-impact papers, filtered your way
-
3Deep-diveMethods, findings and the research gap
-
4MethodologyDesign and step-by-step procedure
-
5OutlineYour table of contents, word budgets
-
6WriteSection by section, cited from your papers
-
7PolishHumanise and reduce source overlap
-
8FiguresFlow diagrams, charts and timelines
-
9ReviewJournal-style critique
-
10Final reportAPA · IEEE · Chicago · Harvard · MLA
Writing the paper is harder than doing the research.
- ✕The blank page paralyses even after months of reading.
- ✕AI tools produce flat, generic prose that any reviewer spots in two paragraphs.
- ✕Citation management is a chore; reformatting between APA, Harvard, MLA, IEEE, and Chicago is worse.
- ✕Generic AI tools hallucinate references, mangle numbers, and forget your sources.
Ten guided steps. Your sources. Your voice.
- ✓Each step does one job — and you can edit, regenerate or refine its output before moving on.
- ✓The writer is grounded in the papers you select — full text where available — not the model's training data.
- ✓The humaniser runs up to five focused passes — including one that breaks the uniform fact density and unescorted jargon that trip AI detectors.
- ✓Citations come only from papers you selected — verified scholarly records, DOIs, BibTeX or your own uploads — never invented. APA 6/7, Harvard, Harvard (AU), MLA 8/9, IEEE, Chicago.
How Hanova AI works
From a research question to a finished, cited paper. The agents do the heavy lifting; you stay the author.
From your research question to a finished paper
Three stages. You bring the question and the sources, you steer every step, you download.
You set the direction, review each step, and download a finished, cited paper.
Five stages from blank page to citation-ready paper
-
Step 01
Sign up and set your topic
Create an account, name your project and describe the research — the topic step turns it into questions and search queries.
01 -
Step 02
Find the right papers
Search leading scholarly databases with your own filters — citation impact, quartile, years, citations — or add papers by DOI, BibTeX or PDF.
02 -
Step 03
Read, find the gap, design the method
Per-paper notes and a literature matrix lead to the research gap; the methodology step designs how you will close it.
03 -
Step 04
Outline, write and polish
Use your own table of contents or a template, draft section by section, then humanise. Edit or refine any step.
04 -
Step 05
Final report and export
Read the reviewer critique, then build the final report — contents, citations, references — as PDF · DOCX · MD · HTML · TXT · LaTeX.
05
What the agents actually do
Define your topic
Give a working title and a short description. The topic step proposes sharper titles, a problem statement, research questions, objectives, keywords and search queries — all editable.
- ✦ Discipline and paper type set the direction
- ✦ Optional target journal
- ✦ Edit anything before moving on
Find high-impact papers
Search leading scholarly databases with your own thresholds. Every result shows its journal, impact score, quartile, citations and a one-line reason it is relevant. Tick the papers you want to build on.
- ✦ journal quartile, citation impact, years, citations, open access
- ✦ Add papers by DOI, BibTeX or PDF upload
- ✦ Open-access PDFs fetched and indexed automatically
Read deeply, find the gap
Each selected paper is read — full text where available — and summarised: problem, method, data, findings, limitations. A literature matrix and synthesis lead to the research gap.
- ✦ Literature matrix across all papers
- ✦ Themes, agreements and contradictions
- ✦ A research-gap statement you can edit
Design, outline and write
The methodology step designs your approach. The outline follows your own table of contents or a paper-type template, and each section is drafted with citations to your selected papers.
- ✦ Notes and word counts per section
- ✦ IMRaD, IEEE conference, review, thesis chapter, proposal
- ✦ Rewrite any single section
Polish, add figures, review
Humanise the prose and reduce overlap with your sources, add flow diagrams and charts, then get a journal-style critique. Select any passage in any step and say how to change it.
- ✦ AI-likeness and overlap score per section
- ✦ Methodology flow diagrams, results charts, PRISMA and timelines
- ✦ Before / after view and version history on every step
Final report and export
Assemble the finished paper: title block, abstract and keywords, table of contents, numbered sections, in-text citations and a reference list of only the papers you cite.
- ✦ APA 6/7, Harvard, Harvard (AU), MLA 8/9, IEEE, Chicago
- ✦ Switch style and reformat in seconds
- ✦ PDF, Word, Markdown, HTML, text or LaTeX
Ten steps. One coherent paper.
Each agent has a single responsibility and a tight prompt. Together they follow how a careful scholar actually works — scoping the question, reading the literature, finding the gap, designing the method, then drafting, revising, seeking critique and only then formatting citations. You approve each step.
Topic
Turns your idea into title options, a problem statement, research questions, objectives, keywords and database search queries.
Literature
Searches scholarly databases, attaches journal impact score and SJR quartile, applies your thresholds, and ranks papers by relevance and quality.
Analyst
Reads each selected paper — full text where available — and extracts problem, method, data, findings and limitations, then synthesises themes and the research gap.
Methodology
Designs the research approach, procedure, data, variables, tools and evaluation, justifying each choice with your selected papers.
Planner
Builds the outline — from your own table of contents or a paper-type template — with word budgets, key points and the papers each section should cite.
Writer
Drafts each section from your notes and retrieved evidence, with [cite:key] markers limited to papers you selected. Banned phrasing filtered.
Humaniser
Up to five focused passes per paragraph, then an originality pass that rewrites anything too close to your sources — meaning kept, expression changed.
Figures
Draws methodology flow diagrams, architecture diagrams, results charts, PRISMA selection diagrams and project timelines — charts use only your own numbers.
Reviewer
Adopts a senior journal reviewer voice. Returns a verdict, major and minor comments, and suggestions.
Final report
Formats in-text citations and the reference list in your chosen style, adds the title block, contents and numbered headings. No model call — deterministic.
Three knobs. Nine settings. Total control.
You decide how aggressive the humanisation is, how much overlap with source material to tolerate, and how creative the language gets. Tune per project.
Humanisation
Drives how many focused rewrite passes the humaniser runs on every paragraph. The naturalise pass — which breaks fact density and escorts jargon — runs at every level.
- Low 2 passes — naturalise + polish
- Medium 3 passes — restructure + naturalise + polish
- High 5 passes — restructure + naturalise + voice + polish + destabilise
Similarity reduction
Sets the 5-gram Jaccard overlap threshold for rewriting source-similar passages.
- Low Only rewrite near-duplicates (≥ 0.85)
- Medium Balanced (≥ 0.65)
- High Rewrite moderate overlap (≥ 0.40)
Creativity
Controls how varied the wording is across writing, humanising and originality passes.
- Low Conservative (0.35) — precise, technical
- Medium Balanced (0.75) — default
- High Bold (1.05) — richer phrasing
Built for every research deliverable
Everything a researcher actually needs
Hanova AI doesn't replace your judgement — it relieves the friction of structuring, drafting, and re-reading. The author is still you.
Questions scholars ask
Will the writing pass an AI detector? +
We deliberately make no such claim. Hanova AI optimises for sentence-rhythm variety and source-overlap reduction, and shows you an AI-likeness and overlap score per section — but AI detectors are noisy and adversarial. Always review and revise before submitting.
Where do my documents go? +
Your documents and projects are private to your account. The passages each step needs are processed by Hanova AI to generate your text, and nothing is shared with other users.
Can I skip citations for a report? +
Yes. Citations only come from the papers you select in the Literature step — select none and the paper is written without citations or a reference list. For your own text, use “Polish my own text” with citations switched off.
What if a paper is not in the search results? +
Add it by DOI, paste its BibTeX (books and websites too), or upload the PDF in the Literature step. For paywalled papers found by search, upload the PDF to give the agents its full text.
How long does each step take? +
Most steps finish in seconds to about a minute: a literature search takes around 30 seconds, drafting a 2,000-word paper well under a minute on the default model, and the final report about two seconds. Higher humanisation and longer papers take longer — and you decide when to move on.
Which AI model writes the draft? +
Hanova AI uses advanced AI language models. You choose the quality per project — Fast, Balanced or Premium — and can switch at any time.
Is there a free plan? +
Yes — during early access Hanova AI is free for individual researchers.
Can I install it on my own server? +
Yes. Universities and research groups that need full control over where their data lives can run Hanova AI on their own servers.
Start your next paper today.
Sign up free, set a research question, and let Hanova AI guide you from the literature to a finished, cited paper — one step at a time.