GlyphMend reconstructs PDFs into structured, inspectable Markdown — locally. It preserves reading order, headings, lists, tables, equations, images, and page boundaries, then exports the same reconstructed document to Markdown or DOCX. Your document stays on your device by default.

Hey Product Hunt 👋

I built GlyphMend because I kept running into the same annoying problem: PDFs look fine to humans, then completely fall apart when you try to turn them into structured content.

Tables become nonsense. Equations disappear. Reading order breaks. Scanned pages need OCR. And sometimes the output looks perfectly plausible while being quietly wrong — which is probably the worst failure mode.

So GlyphMend has a deliberately narrow job:

PDF in → faithful, deterministic, structure-aware Markdown out.

It tries to reconstruct what it can confidently recover, preserve what it can’t, and avoid inventing structure just to produce prettier output.

The browser version runs locally on your device, with OCR, resumable checkpoints, tables, equations, source visuals, Markdown preview/export, and optional DOCX output.

It’s open source, still evolving, and I’d especially love feedback from anyone dealing with difficult PDFs, research papers, technical documents, or document ingestion for AI/RAG systems.

And yes, PDFs have already found several creative new ways to ruin my day while building this. 😅

Quick language update: GlyphMend’s interface is now available in English (default), Russian, and Simplified Chinese. You can switch languages from the app’s language menu. The PDF-to-Markdown workflow is now easier to use in the language you prefer.