# Making a portfolio legible to AI agents

> llms.txt, markdown mirrors of every case study, a schema.org profile endpoint and a For Robots page — how leyra.es became machine-readable without losing its soul.

- Date: 2026-07-17
- Folder: Field Notes
- Tags: llms.txt, SEO, agents, Next.js

Context: this site is a heavily art-directed OS fiction — scroll films, bouncing DVD logos, Win95 popups. Crawlers and AI assistants saw almost none of it. The richest case-study prose literally never rendered.

## Constraint

Nothing on the landing, crafts or the advertising index could change visually. The machine layer had to be additive: metadata, JSON-LD, and new machine-first surfaces.

## Shipped

- llms.txt rewritten to the llmstxt.org format — H1, blockquote pitch, link lists.
- Every advertising case study mirrored as raw markdown at /projects/<slug>/md.
- A schema.org Person endpoint at /api/profile.json, CORS open, for any agent to fetch.
- A human-visible /agents page that tells your agent where everything lives.

## Verdict

The site now has two front doors: one for humans who like bouncing logos, one for the agents that increasingly decide who gets remembered.

---

Canonical: https://leyra.es/notes/making-a-portfolio-legible-to-ai-agents
Author: Fernando Leyra — https://leyra.es
