Import documents, web pages, and source notes

Imports give the project material to work from. They do not prove that the source is accurate, complete, current, or yours to publish.

By AI Ebook Generator editorial teamUpdated August 28, 20267 min read

Use one of four current import paths.

Open a project, go to Author Lab, then choose Sources. You can keep imported material in the source library or turn structured material into editable manuscript chapters.

Current import support and practical limits
InputWhat happensImportant limit
DOCXText and headings are extracted for a saved source or chapter import.The file must be 10 MB or smaller. Review lists, tables, footnotes, and complex layout after extraction.
PDFSelectable text is extracted for a saved source or chapter import.The file must be 10 MB or smaller. A scanned image-only PDF needs OCR before import.
Public URLReadable text is fetched from a publicly reachable web page.Private, sign-in-only, local, and blocked network addresses are not import sources.
Pasted textNotes, research, interview material, or draft sections are saved as a source or imported into chapters.Give the source a useful title and paste at least enough text to identify its purpose.

Keep reference material separate from manuscript text.

Choose Keep as a source when the material should inform drafting, research, or verification. Choose Import as chapters when the document already contains a manuscript structure you want to edit directly.

A source can support the book without appearing word for word in the manuscript. That separation makes it easier to replace weak evidence and avoid accidental copying.

  • Source library: Best for research notes, interviews, examples, reports, and background reading.
  • Import as chapters: Best for an existing draft with useful headings and sections.
  • Full-book notes: Best for concise facts, recurring examples, or constraints that should guide every generated chapter.

Clean the file enough to make extraction predictable.

Use real heading styles in DOCX instead of making every title large and bold. Remove duplicate headers, footers, navigation, and boilerplate when they add noise. For PDFs, confirm that you can select and copy a paragraph before uploading.

For a web page, use the canonical public page rather than a search result, preview link, or share link that needs a sign-in.

Compare the extracted text with the source before drafting.

Check titles, chapter boundaries, omitted text, broken words, tables, quotations, and source links. Extraction favors useful text over visual fidelity, so a document that looks correct can still need cleanup after import.

  • Confirm structure: Rename or reorder imported chapters before generating new material.
  • Record provenance: Keep the source title and original URL when a claim may need verification.
  • Remove private data: Delete personal or confidential material that the project does not need.
  • Check permission: Import only material you are allowed to process and publish from.

Send the project context, not your secrets.

Email the account address, project title, page, and a short description of the issue. Never send a password, reset token, or full card number.

Email support