Reflow: GitHub drops layout optimizer for AI content

If you've ever asked an AI to generate a document, report, or webpage, you already know the output often looks like it was laid out by a drunk person: wonky headings, random spacing, zero visual hierarchy.

Reflow exists to fix this mess. It's an open-source GitHub project that takes content from language models and "reflows" it by applying smart layout rules: fixes headings, paragraphs, margins, lists—everything needed to make text that's technically correct but visually awful actually readable.

The cool part? It works automatically. Feed it raw output from your favorite model (GPT, Claude, Llama, whatever) and it returns something that looks like a human formatted it. No manual prompts to "improve formatting," no copy-pasting into Word to clean up.

Reflow uses a mix of semantic parsing and typographic rules: recognizes text structure, applies visual hierarchy, handles code blocks and lists, and produces clean output in HTML, Markdown, or PDF. You can configure custom styles or use included presets.

The project is still young but the idea is solid: if AI is producing content for humans to read, it should at least present it decently without making you waste time fixing it by hand.