Blog

Measured numbers, real comparisons, and practical workflows for turning documents into AI-ready Markdown.

• How to fix “context window exceeded” in ChatGPT and Claude 7 August 2026

  Your file isn’t too big — it’s too expensive in tokens, and most of them are avoidable. The fix, in order of impact.

• PDF to Markdown for ChatGPT & Claude: the measured token savings 7 August 2026

  Everyone repeats “30–60% fewer tokens.” We measured it per format — including the one where Markdown loses.

• MarkItDown vs Docling vs Marker vs Pandoc: an honest comparison 7 August 2026

  Four engines, four different jobs — including the cases where the engine we build on is the wrong choice.

• Convert documents to Markdown entirely offline on your Mac 7 August 2026

  Who actually needs local conversion, what “local” has to mean, and how to verify any converter’s claims yourself.

• Preparing documents for RAG: clean Markdown and heading-aware chunks 7 August 2026

  Retrieval quality is decided before the chunker runs. A practical pre-processing pipeline, start to finish.

• Batch-convert a folder of Word and PDF files to Markdown 7 August 2026

  Drag-and-drop batches, watch folders, the CLI, and Shortcuts — four ways to convert at folder scale, all local.

• Import Word docs and PDFs into Obsidian as clean Markdown 7 August 2026

  Turn trapped documents into real, linkable notes — plus an inbox folder that converts itself into your vault.