How this newsletter is built

This page is about how robertomb is made: why I built it instead of using a newsletter platform, and what it runs on.

Why I built it

I wanted a newsletter that felt like my own place on the internet. A platform would have been faster to start with, but I wanted the domain, the archive and the subscriber list to be mine, and I wanted to understand every piece of the system.

I also had a few rules for it:

• Light and fast. Pages are plain HTML and load almost instantly, even on a slow connection.
• Easy to keep running. No database and no admin panel to maintain. Posts are text files.
• Easy to find. Every issue has a permanent, public page that search engines and AI assistants can read and link to.
• Readable everywhere. In the browser, in RSS readers, and in any email client.

It also helps that I work on growth at Resend. Running my own newsletter on it is the best way I know to feel what our users feel.

How it works

1. I write each issue as a Markdown file and commit it to Git.
2. The site rebuilds into static pages: one per issue, plus the archive, the RSS feed and a few files for search engines.
3. When an issue is ready, I send myself a test email, read it on my phone, and then send it to everyone with a single command.

That's the whole workflow.

What it's built with

Email: Resend.

• Broadcasts send each issue to the whole list, with tracking for opens and clicks.
• Contacts and Segments hold the subscriber list, so there's no database of my own.
• Unsubscribe links are handled by Resend in every email.
• React Email renders the email templates. Each issue is laid out for email clients, not copied from the website.
• Double opt-in: when you subscribe, you get a confirmation email with a signed link that expires after 48 hours. You're only added once you click it.

Site: Next.js on Vercel.

• Every page is generated in advance, so there's no server rendering to wait for.
• Fonts come from your device, so there's nothing extra to download.
• Light and dark mode follow your system setting.
• Privacy-friendly analytics from Vercel, with no cookies.

Discoverability.

• Structured data on every page describes each post and its author for search engines.
• /llms.txt and a Markdown version of every post (add .md to any post URL) make the archive easy for AI assistants to read and cite.
• A sitemap, a full-text RSS feed, and new posts submitted to search engines through IndexNow.

Want one like it?

I plan to open-source the whole project so anyone can run a newsletter like this with their own domain. Until then, if you'd like to set up something similar, email me. I'm happy to share what I learned.