Science deserves to be seen.

What is PublicationGod?

PublicationGod empowers researchers to bring valuable yet unpublished data to light — data that often ends up in the file drawer after a thesis, pilot project, or smaller study. Too many scientifically sound results never make it to publication simply because they are seen as low-impact, negative, or not worth the time to write up. Yet every robust dataset contributes to the collective progress of science.

Our mission is to make it easy and rewarding to share such work. Using AI-assisted and guided structured drafting, PublicationGod helps researchers transform figures into the first draft of a publication-ready manuscript within hours, not months.

By dramatically reducing the time and cognitive load between solid data and a coherent scientific draft, PublicationGod lowers the threshold for sharing knowledge that would otherwise remain unseen. We believe that transparency and completeness in scientific reporting strengthen the foundation of all research — and that even small, "unimpactful" findings deserve their voice in the literature.

PublicationGod was developed by researchers at the University of Lübeck, Germany, and is operated by Mittag Advanced Writing Technologies UG (haftungsbeschränkt). Every dataset that meets scientific standards deserves to be published. PublicationGod exists to make that happen — faster, ethically, and collaboratively.

Is it ethical to use AI for scientific writing?

The ethically decisive step in science lies in the generation and presentation of robust experimental data — a responsibility that remains entirely with each individual researcher. All users are required to confirm adherence to the principles of good scientific practice upon registration.

AI-assisted manuscript drafting does not replace the researcher's integrity or expertise. Instead, it helps transform robust data rapidly into a clear, well-structured, and shareable manuscript — turning results into knowledge. With PublicationGod, this is a stepwise process, and each step needs to be individually confirmed. The AI does not simply generate a full paper with a single click. Instead, we assist and accelerate the process from data to manuscript — but the authors remain in full control of the generated output.

In the end, what matters most is not the path taken to write the text, but the transparency and reliability of the science behind it. Most major organizations explicitly allow the use of AI in scientific writing, including COPE (Committee on Publication Ethics), the DFG, and the European Code of Conduct for Research Integrity (ALLEA, 2023) — but all require transparent declaration. At PublicationGod, this happens automatically at the export stage, when a corresponding paragraph is added to the methods section.

Moreover, by enabling direct upload to the open preprint server Zenodo with a single click, we aim to support Open Science — so that valuable results, even small or unexpected ones, can see the light of day and contribute to scientific progress.

Why PublicationGod instead of general AI tools?

General-purpose AI chatbots are not designed for structured academic publishing workflows. References are occasionally hallucinated and require manual verification, figures must be interpreted separately, and manuscript formatting remains fragmented across multiple platforms. Other subscription-based AI writing tools typically still require external software for reference management, figure handling, or preprint submission — and often use US-based servers.

PublicationGod integrates PubMed-based literature retrieval, structured manuscript drafting, figure analysis, collaborative editing with invited co-authors, and one-click preprint upload with citable DOI within one academic workspace. Instead of a running monthly subscription, you pay per manuscript only when needed. Your data stays on EU servers, and the entire workflow remains in one workspace.

Is this not just fostering papermills?

"Papermills" typically recycle the same manuscript structure over and over — changing only superficial details, such as swapping one microRNA for another or replacing a cancer type while keeping most of the text identical. Such practices undermine scientific integrity and flood the literature with unreliable work.

PublicationGod is fundamentally different. It is not a shortcut for mass-producing manuscripts, but a guided, structured workflow that requires meaningful input at every stage — from defining the study logic to uploading validated figures and datasets. Each project must be built from real data, and every section is generated interactively. The token-based system further ensures that automated, repetitive production would be neither practical nor cost-effective.

In short, if someone wanted to run a papermill, a word processor with "search and replace" would sadly be faster and cheaper. PublicationGod exists to do the opposite — to empower honest researchers and make real science visible.

How does PublicationGod work?

PublicationGod guides you through the complete manuscript writing process in clearly defined steps — from your first figure to a submitted preprint. There is no blank page, no guesswork about structure, and no risk of losing work between tools. The full workflow typically takes between 25 and 60 minutes for a first draft.

Here is what happens at each stage:

1. Upload your figures - PublicationGod analyzes each figure using AI and generates a structured figure legend as a starting point. You review, edit, and confirm before moving on. Nothing is hidden from you.
2. Build your reference library - You search PubMed directly within the platform and add the references relevant to your work. PublicationGod only draws from this curated library — it never invents citations or halluccinates references. Your literature, your paper.
3. Draft each section — step by step - The workflow guides you through Results, Methods, Discussion, Abstract, and Introduction in a logical order. For each section, AI generates a first draft based on your figures and your library. You confirm, edit, and refine each output before proceeding.
4. Invite co-authors - Once a draft exists, you can invite co-authors directly into the project. Everyone works on the same document, edits are tracked, and no version gets lost in email chains.
5. Export or submit as preprint - When you are satisfied with the draft, export it as a Word document or LaTeX file — or submit directly to Zenodo with one click to receive a citable DOI. The COPE-required AI disclosure statement is added automatically to the methods section at this stage.

Throughout the entire process, PublicationGod operates on EU servers — your data is never used to train AI models, never leaves the EU, and is processed in full compliance with GDPR. The token-based payment model means you pay only for what you actually use, with no monthly subscription and no lock-in.

• Guided and structuredEvery section follows a defined academic workflow. You are never staring at a blank page wondering what comes next.
• No hallucinated referencesPublicationGod only cites from the library you build. Citations are verifiable, traceable, and always under your control.
• COPE-compliant by designThe AI disclosure statement required by COPE, DFG, and ALLEA is added automatically at export — no last-minute edits, no policy risk.
• EU servers · GDPR compliantAll data is processed and stored on EU infrastructure. Your research data is never used to train AI models.
• No subscription · Pay per useBuy tokens only when you need them. One token package is typically sufficient for a complete small manuscript. Invoice payment available for universities and institutions.

The tutorial video below walks through the complete workflow in real time — from uploading the first figure to submitting a preprint to Zenodo (24 minutes). Every step is explained in detail, including the reasoning behind each stage of the process.