Services & Features
Everything MarkOut does — free, private, and entirely inside your browser tab.
Highlight extraction
A custom YOLO26 model detects highlighted areas visually on every page — works with highlights made in Acrobat, Preview, Edge or any other reader, even if the PDF doesn't keep the annotations.
Scanned PDF support
The same neural pipeline handles image-only PDFs and screenshots: one flow for every document, digital or scanned.
OCR in the browser
PaddleOCR recognizes the text of every fragment, self-hosted and offline-capable — no text ever leaves your machine.
Clean Markdown export
Highlights come out in reading order, grouped by page and section, with YAML front-matter and stable citeable IDs.
LLM-ready output
The exported Markdown is designed as a base for AI assistants: stable structure, per-line IDs, extraction-quality notes.
100% private
No upload, no account, no server. Processing happens locally; at the end the models and file are released from memory.
Who it's for
- Students — turn highlighted papers into revision notes in seconds.
- Researchers — build a clean corpus of passages from many PDFs, ready for analysis or LLM workflows.
- Writers & editors — collect quotes and references without retyping.
- Privacy-conscious users — any document, sensitive or not, stays on your device.
Platforms
MarkOut runs on any modern browser with WebAssembly: Chrome, Firefox, Edge, Safari 16+. Desktop gets optional WebGPU acceleration; mobile uses a memory-optimized pipeline. No installation, no account, no plugins.