Show HN: Scribe – generate and preview transactional emails from plain English

usescribe.ashpak.dev

1 points by blackmamoth 2 days ago

I built Scribe, a tool for generating styled transactional emails by describing what you want in plain English, with an instant live preview.

The preview runs via client-side Babel compilation, so changes show up immediately without a server roundtrip.

Scribe uses React Email and Resend, which helps ensure the generated emails render consistently across Gmail, Outlook, Proton Mail, and other major clients.

Features include:

- Live preview while iterating - Brand management (save colors/logos, switch between brands) - Version history with rollback - Send test emails before exporting - Export email code

The project is open source and doesn’t require signup.

I’d love feedback on what would make this genuinely useful in production vs just a demo.

https://usescribe.ashpak.dev https://github.com/Blackmamoth/scribe

blackmamoth an hour ago

I wanted to build my own v0-style code generation tool, but starting with a full web app builder felt overwhelming. So I narrowed the scope to something more focused: an AI email generator.

What it does: Describe an email in plain English, select your brand (logo, colors), and get a responsive, styled email. The AI generates React Email components, which means you get email-client-compatible code that you can version control and customize. Export as React or HTML.

One interesting technical decision: Initially used WebContainer for live preview, but boot time was 30-40 seconds plus 3-4s compilation. Switched to client-side Babel transpilation - now previews render in <1s. Trade-off is you can't install additional packages (e.g., custom icon libraries), but React Email's built-in components cover most use cases. Did I make the right call here?

Stack: TanStack Start, React Email, Postgres, Vercel AI SDK, Resend

Try it: https://usescribe.ashpak.dev GitHub: https://github.com/Blackmamoth/scribe

Would love feedback on whether this solves a real problem or if existing tools are already good enough.