I installed EntityMap on this site.

This morning I shipped a file to my site that nothing can read yet. That was the point.

EntityMap hit v1.0 on 27 May 2026. The idea is small and obvious once you hear it. Sitemap.xml tells crawlers which URLs exist. EntityMap tells AI systems which entities exist, how they relate, and where the evidence for each one lives. Same instinct, one floor up. Mine is sitting at /entitymap.json now, with a readable companion at /entitymap/ and a line in the footer.

Here is the uncomfortable bit. A day in, the number of AI systems reading it is zero. That includes ChatGPT, Claude, Perplexity, and Google. The GitHub repo has 11 stars. Waikay’s own generator still says “Coming Soon”. Measured today, the file does nothing. Search Console looks exactly like yesterday and the AI Overview numbers have not twitched. This is Wave 0, and I am the wave.

The case for shipping it anyway comes down to 2 things.

First, the names on it. Dixon Jones is one of 2 editors. He founded Majestic, runs InLinks, and has spent over a decade on entity-based search. When someone with that record puts his name on a brand new spec, it earns a place on my watch list on credibility alone. His Waikay co-founder Fred Laurent is the other editor. They published under CC BY 4.0, so it stays open even though Waikay will eventually sell the tooling around it.

Second, the maths. The whole thing cost me half a day. 2 hours to write an entitymap.json by hand, declaring 9 entities, me, my 2 companies, and the 6 topics I want my name stuck to (AI agent SEO, schema for LLMs, MCP, llms.txt, Markdown for Agents, and Search Journey Optimisation). Another hour for the HTML companion. The footer was 1 line. That is the entire cost. The payoff, if EntityMap catches on, is that I was already there.

I have watched this film before. Schema markup took most of a decade to earn out. Sitemap.xml took years to be taken seriously. Most protocol-layer work is registration, not advertising. Ship EntityMap expecting citations and you will be irritated by Thursday. Ship it as one more piece of entity infrastructure that might quietly compound while you forget it exists, and the bet is sensible. That is the same argument I made in The Three Lives of Schema Markup.

Two things if you fancy trying it. Write the JSON by hand, or with a generator that is not tied to one company’s product. Wiring your entity declarations into a single vendor’s workflow is the sort of quiet decision you regret 2 years later. And be ruthless. EntityMap is not a content index. Declare the few entities you actually want to own, not every post you have published. Tight beats complete.

The agent-ready post stays as it is. EntityMap joins the protocol-stack list there the day a real consumer, a major LLM, a Chrome Lighthouse check, a Cloudflare default, actually acknowledges it. Until then it lives here, in my footer, and on my entity map page, holding its position.

Watching, not betting.