A web app for academic publishers that turns Word manuscripts into journal-ready PDF, HTML galley, JATS XML and Crossref files in minutes, with a live editor, author proofs and each journal house style built in.
Most journal publishers still format every accepted article by hand: the text is copied from Word into a layout program, headings, tables, equations and references are fixed one by one, and the XML for indexing is made separately. One article can take an hour or more, and every new issue means days of layout work.
ManuscriptFlow does this in the browser. The editor drops in Word files, as many as they like. The app reads the title page, authors and affiliations, abstract, keywords, headings, tables, figures, equations and references, applies the journal house style and typesets a finished PDF with the journal cover, an article information box, running headers and the DOI in seconds. The same click makes the HTML galley and JATS XML for OJS and the Crossref deposit file.
Editors correct anything in a Word-like document view next to a live PDF preview, send authors a private proof link where they edit or comment, and accept only the changes they want. It runs as a multi-tenant SaaS: every publisher has its own workspace with its journals, covers, rules and users, and publishers that still finish pages in InDesign can download an IDML file built on their own template.
Accepted articles arrive as Word files written in every possible way: headings typed in capitals or numbered, several affiliations on one line, tables with merged cells, equations from different editors, references in mixed styles. Formatting them by hand takes a skilled designer, the result depends on who did it, and the indexing files are a separate job. The software had to read messy manuscripts reliably, produce pages a publisher is happy to print under its own name, and still let a person fix anything in a few clicks.
We built a Python web app around one model of the article. A reader takes the Word file apart down to text runs, tables, equations and pictures; a classifier finds the title page, sections, captions, notes and references and renames headings to the journal standard; house rules then clean the text the way a copy editor would. A built-in typesetting engine lays the article out in two columns with tables and figures placed as floats, the journal cover and the article information box, with no desktop software. The same model is written out as PDF, HTML, JATS XML, Crossref XML and InDesign IDML, so every output says the same thing. Author proofs work on a copy of the article, and the changes are compared word by word so the publisher can accept them one at a time.
A multi-tenant platform that gives every business its own AI agent — answering customers on the website, Messenger and...
A managed website-care service with its own client dashboard: round-the-clock uptime monitoring, nightly health checks...
A multi-tenant SaaS platform that works like a marketing hire rather than a content tool. It learns the brand, plans t...
Every project on this page started with a conversation. Yours can too.
