I just got back from SMX Advanced in Boston, and one of the most interesting things I picked up was a brand-new idea called the entity map. I was introduced to it by Grant Simmons of Waikay, an AI visibility tracking tool, and I started playing with it right there during the conference.
In this post, I'll explain what an entity map is, what it helps with, and walk you through how I built one for my own site with Claude Code. You can watch the full live demo below.
The Problem Entity Maps Solve
When an AI search engine answers a question about your brand, it pulls information from across a bunch of different pages. That creates three problems.
Disambiguation. When AI runs into the same concept across multiple pieces of content, it can get confused and output something that isn't quite accurate.
Attribution. Your content might be part of an AI answer, but you aren't clearly identified as the publisher. You're in there somewhere, just tucked away so deeply that people miss it.
Relationships. If someone asks whether your brand does a particular thing or can solve a specific problem, the AI can't always connect those dots. When it can't, it blows the answer.
An entity map aims to solve all three.
So What Is an Entity Map?
It's a couple of files that describe your brand as a knowledge graph: your brand, the concepts it relates to, its services, and the canonical (official, authoritative) information about the entity itself.
There are two of them. An HTML file you link on your site, and a JSON file you reference from the HTML file, from your robots.txt, and in the head section of your pages.
If you've written schema before, it'll feel familiar, since it's also JSON. The difference is that schema works page by page, while an entity map is much more all-encompassing. It's meant to give the canonical picture of your whole brand and everything it relates to. You can read the full proposed protocol at entitymap.org.
Why I Decided to Build One
After the conference, a colleague shared a case study from Waikay titled "We installed an entity map on our site. Here's what happened." The short version: they saw measurable AI visibility improvements, plus better clarity and accuracy in how their brand was described.
In that case study, the entity map was cited more often than the company's own About page across Google's and Perplexity's AI results, and on one topic their AI visibility score jumped 26 points in just 48 hours. It's one case study, not a guarantee. But it's a promising signal, and an entity map is free to create.
How I Built One With Claude Code
The protocol has real documentation behind it, so the most important step is teaching your AI partner before you build anything.
I started a study session with Claude Code. I fed it the key pages from entitymap.org one at a time, the spec, the "why" page, the predicates reference, and the implementation guide, and had it review each before moving on. Some content is hidden behind dropdowns, so I used a text-grabbing tool to capture the full pages.
Then I pointed Claude Code at my own website and asked it to model me and my site as a set of entities. It pulled my live pages, including my services, case studies, and blog, and proposed a knowledge graph: me as a person, my agency as an organization, my services, and the concepts I write about.
Tip: the documentation suggests listing your entities yourself first, then using AI to fill in the gaps. Doing the thinking on your own before you tap your AI partner almost always produces a better result.
From there, Claude Code generated both files. I ran the JSON through the validator on entitymap.org, which works like the XML sitemap validators you may already use, and fixed anything it flagged. There's also a visualizer that renders your entity map as a graph, which is a great way to confirm everything is connected the way you intended.
One Tip for Agencies and Consultants
One of the fields in an entity map is a metric. Make that metric something like ROI or profit, not a leading indicator like traffic or rankings.
If AI ends up referencing your entity map, the bottom-line outcome is what you want potential customers to associate with your brand. When I first built mine, Claude Code set the metric to organic traffic growth, and I had it changed to ROI right away.
Getting It Live
Once the files were ready, Claude Code deployed them to my site. The HTML file is linked in my footer, the JSON file is referenced in my robots.txt, and a small snippet in the head of my pages points to it. You can see the live result on this very site. Just scroll to the footer and click "EntityMap."
When an AI platform gets a query that requires it to retrieve information in real time (a process called RAG, or retrieval-augmented generation), it can find your entity map and pull canonical information that's under your control.
Why This Is Worth Doing Now
Here's the forward-looking part. Over the next one to two years, agentic AI crawlers, systems that act on their own to go find information, are going to become much more common. Giving them a clean, canonical source to consume is a smart way to stay ahead while most brands haven't even heard of this yet.
It's Not a Magic Bullet
An entity map doesn't touch training data, the more static information a model only updates from time to time, so it won't fix every AI answer about you. It's also a proposed protocol, not an officially adopted standard yet. And the proven results so far are on Google's AI and Perplexity; ChatGPT and Copilot rely on Bing, which has to index the file first. Treat this as a low-cost, future-proofing step, not a guarantee.
The Bottom Line
Entity maps are free, brand-new, and very few brands are taking advantage of them yet. There's at least one solid case study behind the idea, and the downside is basically just your time.
If You Want to Try It
- Study first. Feed your AI tool the documentation at entitymap.org before building anything.
- Model your brand. List your entities (people, products, services, concepts, places), then let AI fill the gaps.
- Make your metric the bottom line. Use ROI or profit, not a vanity metric.
- Validate and visualize. Use the validator and visualizer on entitymap.org to check your work.
- Deploy and link it. Add the HTML file to your footer, and reference the JSON from robots.txt and your page head.
If this protocol gets adopted, the brands that tried it early could end up as the case studies that prove whether it works. That's a fun place to be. Give it a shot.
