All work

Publishing pipeline

Tools that let a small publisher typeset manuscripts automatically, output e-books and print PDFs and prepare them for distribution.

Status
Prototype
Technology
Python, Node.js, Markdown, EPUB3, DOCX

The situation

When importing from Word, common typesetting programs lose headings and whole sentences. The formats are binary and can be neither automated nor versioned cleanly.

The idea

One single plain-text source from which e-book and print PDF are produced. An agent only formats in small, reversible steps: plan, compare, apply, undo. A person checks along the way.

How it is built

  • Prototype for EPUB3 and a 6 × 9 inch print PDF with an embedded free typeface and real footnotes.
  • Every e-book is checked automatically with EPUBCheck.
  • Export to Word and generation of typesetting files, each with tests.
  • For one manuscript, the text lost in the Word import was fully reconstructed.

Documented figures

  • 160subheadings restored from a Word import, none lost (June 2026)
  • 0errors and warnings in EPUBCheck (June 2026)

Status

Prototype. Next step: making the pipeline ready for translations into Spanish and Portuguese.

A similar problem?

Tell us in one sentence what it is about. We will tell you honestly whether we can help.

Write an email
Next projectCompanion app with AI chat

Assistant

Fixed answers, no AI