Getting Started

1. Create a project

Sign up, verify your email, and you land on your Dashboard. That is the list of your projects. One project is one manuscript. Press New project and you get a two-step form.

1 Choose how you want to start. The three options are described below. Nothing else appears until you pick one, and you can go back with "Choose a different way to start" without losing what you typed.
2 Fill in the details. A project name, a journal, and (on the upload paths) your file.

The project name is not your manuscript title. It is the label on the dashboard card, up to 100 characters, so make it something you will recognise ("Nanophotonics paper, IEEE resubmission" beats "Paper 3"). The real title goes in the Metadata tab.

The journal is a setting, not a commitment. Change it later in Project Settings and regenerate. Your content never has to change.

2. Choosing the journal

The picker works in two steps: choose a publisher, then start typing the journal name. Search matches the journal's name and its common alternative names, and results are grouped by imprint.

Some journals offer more than one format (a single-column review version and a double-column final version, for example). When they do, you get a short list to choose from with the usual one already selected.
If your journal genuinely is not there, pick "My journal isn't listed". You get that publisher's generic format, which is close enough to write and review in, and we are told which journal you needed so we can add it. You can switch to the real one once it lands.
Not aiming at a journal yet? TeXpress, Plain Article is a neutral single-column format that always compiles. It is a good place to work while you decide.
Journals differ in how deep their headings go. If you switch to a journal that allows fewer levels than you have used, you are warned first and told exactly what happens. See Troubleshooting.

Missing a journal you need often? Request it here.

3. The three ways to start

All three end in the same place: your content sitting in the editor tabs. They differ only in how it gets there.

edit_note

Start in the visual editor

A blank project. No file, no AI, no quota spent. You get empty tabs and you write straight into them.

Choose this if you are writing a new paper, or your draft is short enough that pasting it in section by section is faster than correcting an extraction. It is also the honest answer when a document is messy: a clean start beats fixing a bad parse.

upload_file

Upload your Word or PDF

A .docx or a .pdf, up to 25 MB. We read the file, then AI turns it into structured content: title, authors, affiliations, headings and their nesting, body text, figure images and captions, tables, equations, and the reference list.

Word gives better results than PDF, and it is not close. A .docx still carries real heading styles, so we read your structure rather than guessing it from font sizes and page geometry. Use the Word file if you have one.

Choose this if you have a finished or half-finished manuscript and want it formatted for a journal.

code

Upload an existing .tex

A single .tex file. We read the LaTeX you already have and pull it apart into the same tabs. Structural commands (\section, \begin{figure}, \cite, math environments) come through well.

Two honest limits. Macros you defined yourself cannot always be interpreted, so check anything unusual. And figure images are not in a .tex file, only references to filenames, so every figure comes through as an empty card waiting for you to upload the image on the Figures tab.

Choose this if you are switching journals or resubmitting after a rejection and do not want to reformat by hand.

Bring your .bib along

On either upload path there is an optional .bib field. Use it. A BibTeX file is loaded exactly as written, with no AI in between, which makes it the most accurate references you can get. You can also upload one later from the References tab. Forgetting it costs nothing: we will read the reference list out of your document instead, you will just want to check it more carefully.

4. What happens after an upload

You are taken to a progress screen while the extraction runs. It usually takes a couple of minutes; a long, figure-heavy paper takes longer. You can leave the page and come back, the work continues. When it finishes you land in the editor with a banner reminding you to review everything.

Read that banner seriously

Extraction is a very good first draft of your structure, not a finished job. It is normal for it to be nearly perfect on headings and paragraphs, good on tables and equations, and occasionally wrong about which line was a caption. Nothing about the extraction is destructive: every field is yours to correct, and correcting it is faster than typing it.

Usually right

Section headings and how they nest
Body text, bold and italics, lists
Title, abstract, keywords
Author names and affiliation numbering
Equations written with Word's equation editor

Check these first

Figure captions, and that every figure has its image
Table grids, especially merged cells and borderless tables
Which author is the corresponding one
Reference details, if you did not upload a .bib
Anything that was in a text box or a footnote
That every displayed equation actually arrived. Equations set as pictures, or built with your word processor's own equation tool, are the ones most likely to be missed

Limits worth knowing before you upload

An extraction is the one part of TeXpress that costs AI time, so it is metered. Everything else, editing, generating, compiling and downloading, is unmetered.

Your plan sets a page cap per document and a number of extractions per month. Both are shown on your Profile page, with a list of what used them.
The check happens before the project is created. A file that is too long, or one we cannot read at all, leaves you on the form with a plain explanation and no half-made project to clean up.
Starting blank costs no extraction at all, and neither does editing an existing project.
Exact numbers per plan are on the pricing page.