Guides to converting documents into Markdown

A PDF stores coordinates, not structure. What a PDF to Markdown conversion really returns, which documents break it, and how to clean the output up before RAG.

Word keeps headings, lists and tables as real structure, so DOCX to Markdown is far cleaner than PDF. Measured output, the empty table header row, dropped comments and tracked changes, and a repair pass before the text reaches a model.

One sheet becomes one heading and one pipe table. Measured output for Excel files: row 1 promoted to header, NaN from empty rows and merged cells, formulas without a cached value, currency and leading zeros stripped — plus a cleanup script.

Splitting a book into chapters, dealing with illustrations and DRM, and why Markdown is a better base for notes, summaries and training data.

Slide titles, bullets, speaker notes and tables: how to get a readable outline out of a deck instead of a dump of slide fragments.

Stripping navigation and ads while keeping headings, tables and links intact, for documentation, articles and datasets.

Semicolon separators, the BOM that Excel writes, encoding issues and pipes inside values: what breaks a CSV to Markdown conversion and how to fix it.

An array of objects becomes a table, nested values become a JSON string in a cell, and YAML or TOML stays a fenced code block. Limits and workarounds.

Why Markdown beats PDF and DOCX for retrieval: headings as chunk boundaries, tables that stay whole, metadata. A practical checklist for preparing a corpus.

Preparing documents for an LLM: which formats to convert, why Markdown works better than raw text, and how to wire conversion into a pipeline over the API.

Why contracts, medical records and internal documents should not go through cloud converters, and how to convert them without handing the file to a third party.

24 formats in one converter: PDF, DOCX, PPTX, XLSX, EPUB, HTML, JSON, YAML and more. No sign-up, and files are not sent to third-party clouds.