Vesta Press: Read Anything, Beautifully

Why I built it

I read a lot online and most of it arrives ugly: cramped columns, popovers, ads fighting the words. Vesta Press was an attempt at a calmer alternative. Paste a link, drop some text, or drag in a screenshot, and it strips the noise and typesets what matters.

What I explored

• One input, three kinds of source. A link, raw text, or an image. It pulls the real content out and leaves the clutter behind.
• Four ways to read. The same piece can be set as Editorial (magazine-grade serif with drop caps), Minimal (clean sans, generous whitespace), Manuscript, or Document. Reading is a mood, so the typography should bend to it.
• A library and a journal. Somewhere to keep what you've saved, and a space to write alongside it.
• OCR. Point it at an image of text and pull the words out, so a photo of a page becomes something readable.

How it's built

Next.js and React with TypeScript. Article extraction uses Readability and an HTML-to-markdown pass; the Anthropic API handles the AI assistance; Supabase stores everything. There's a clean export-to-markdown path, so nothing you save is trapped.

The whole point was restraint: take something messy and give it room to breathe.