Relaunching Robots First

When I rebuilt my personal website, I started with a controversial premise: the most important visitors aren’t human anymore. AI agents, search crawlers, and LLMs are increasingly how people discover and interact with online content. So I optimized for them first.

The Approach

I wanted to make my site as easy as possible for AI to understand, navigate, and reference. This meant going beyond traditional SEO and thinking about how an LLM actually processes web content.

What I Built

Machine-first architecture: Clean semantic HTML with proper heading hierarchy, complemented by JSON-LD structured data, a comprehensive sitemap.xml, and an llms.txt file that explicitly tells AI agents what’s on the site.

Performance as accessibility: Impeccable Lighthouse scores with minimal JavaScript. Fast load times aren’t just good UX—they make automated navigation trivial.

Dual-format content: Every page available in both HTML and markdown. The markdown versions are pure signal, no noise—perfect for AI consumption while humans get the styled experience.

Explicit over implicit: All images captioned, clear contact paths, and content written with strong semantic structure that doesn’t rely on visual context.

The stunt: I even published a press release announcing the relaunch. Will it get picked up by AI training data? Who knows. But it’s an experiment in thinking differently about distribution.

The Reality Check

Is publishing your entire site in markdown necessary? Probably not for most people—it’s partly a statement piece to make people question their assumptions about audience. But the underlying principle holds: if your content can’t be easily parsed by AI, you’re limiting your reach.

I didn’t get everything perfect. I had a deadline, and the landscape is changing faster than any best practices document can keep up. But here’s the thing: it’s probably better than your site right now.

Want to build something better? Reach out: contact@mirzu.com

---

The Meta Point

This case study itself follows its own principles—clearly structured, scannable headings, explicit takeaways. Because whether you’re human or AI, good information architecture just works.