The Open Standard That Gives AI Systems A Structured View Of Your Business

AI techniques at the moment are answering questions on your enterprise. The issue is that they’re usually getting it incorrect.

Think about the standard scenario. A model’s merchandise, providers, experience, areas, management, and relationships are distributed throughout dozens of pages. An AI mannequin retrieves fragments from these pages, stitches them collectively probabilistically, and generates a solution. The result’s usually hallucinated product names, invented executives, misquoted capabilities, and weak or absent attribution.

This isn’t a failure of AI fashions. It’s a failure of the medium itself. We have now constructed the online round pages, hyperlinks, and prose. AI retrieval techniques want one thing basically completely different: a structured layer of which means and proof.

The Proposal: EntityMap

EntityMap has simply entered public session. It’s a new open commonplace that provides organizations a option to publish a single structured file. This file declares what the group is aware of, maps how its key entities relate to at least one one other, and hyperlinks each declare again to its supply proof.

The session runs till 30 June 2026, with formal launch scheduled for July 1. For the following 33 days, the challenge is actively looking for implementation suggestions, technical critique, and real-world testing from builders, web optimization professionals, publishers, structured-data specialists, and anybody constructing or counting on AI retrieval techniques.

The place EntityMap Sits In The Requirements Panorama

EntityMap just isn’t a alternative for present internet requirements. It fills a niche that sitemap.xml and schema.org had been by no means designed to handle.

Sitemap.xml tells crawlers which pages exist on a web site. Schema.org describes what seems on particular person pages. EntityMap tells AI techniques what a company is, what it is aware of, and the way that data connects throughout all the web site.

This distinction issues. Think about a healthcare group publishing therapy protocols. With schema.org, you can annotate a single page. With EntityMap, you’ll be able to say the next: “Listed below are our core therapy areas. These are the relationships between them. Right here is the peer-reviewed proof supporting every declare. Right here is the place that proof lives on our website.” An AI system studying that file will get a structured view of institutional data reasonably than reconstructing it from web page fragments.

Or, think about a SaaS firm involved about how AI techniques describe its product. EntityMap permits the corporate to declare: “We provide function X. It differs from opponents in Y. Right here is the proof: hyperlink to documentation, hyperlink to case research, hyperlink to comparability web page.” Not should the corporate depend on an LLM to deduce differentiation from scattered internet content material.

The identical logic applies to publishers defending attribution, authorized companies clarifying experience boundaries, monetary providers companies navigating regulatory nuance, and types involved about AI misrepresentation.

How EntityMap Works

EntityMap is a JSON file printed at a predictable location on a site. It incorporates three core parts.

Entities are named issues the group covers: merchandise, providers, folks, ideas, areas, rules, areas of experience.

Relations map how these entities join. Examples: “this product improves this final result,” “this particular person leads this staff,” “this regulation governs this service.”

Proof chunks are supporting passages from the web site, linked to their supply URL.

Every chunk carries attribution metadata: the writer title, the supply web page, the retrieval timestamp. This metadata survives extraction, aggregation, and storage in vector databases. When an AI system generates a response utilizing your content material, the chain of proof stays intact.

The specification is intentionally minimal. The conformance flooring consists of roughly 12 required fields throughout three objects. All the things else is optionally available enrichment: customized predicates, cross-shard decision, verification standing declarations, changelog monitoring.

Who Ought to Pay Consideration

In case you are constructing Retrieval Augmented Generation (RAG) systems, cleaner supply knowledge means higher reasoning chains and fewer hallucinations.

In case you are an web optimization skilled, this represents a new lever for AI visibility. It really works with conventional content material and hyperlink methods reasonably than changing them.

In case you are a writer, this can be a option to declare what you realize and preserve attribution as your content material will get disaggregated throughout AI platforms.

In case you are involved about how AI systems represent your organization, this can be a instrument to claim management.

The usual is printed beneath CC BY 4.0. There isn’t a vendor lock-in, no subscription, no proprietary software program requirement. Group contribution is open. The supply code, specification, and validation instruments are all out there at GitHub.

What The Mission Wants From You

The session interval just isn’t ceremonial. The challenge staff is actively looking for particular types of suggestions.

Technical implementation suggestions: Have you ever tried constructing an EntityMap to your website or product? What broke? What felt awkward in follow?

Use-case validation: Does this clear up an issue you truly face? Does it miss one thing vital to your area or business?

Predicate critique: The usual defines 24 core predicates (IMPROVES, DEPENDS_ON, MEASURES, and others). Are these the proper semantic abstractions to your work? Ought to we add or take away from this listing?

Integration concepts: Are you constructing a generator? A validator? A dashboard for managing EntityMaps? The challenge desires to know what tooling you might be contemplating.

Sector-specific functions: In the event you work in healthcare, finance, schooling, authorized, or one other vertical, what would an EntityMap profile to your sector appear to be?

The specification is out there at entitymap.org/spec/v1.0. A validator is reside at entitymap.org/validate. The group discussion board and GitHub repository are at github.com/entitymap.

Members are invited to overview the specification, check implementation, increase points, counsel enhancements, and contribute to the dialogue earlier than 30 June 2026.

Necessary Context: This Is Genuinely Open

This can be a requirements proposal from throughout the search and AI group. R.V. Guha, one of many founders of schema.org, has reviewed the challenge and given it his endorsement.

The session is genuinely open. The primary section focuses on technical overview and early implementation. Wider adoption, sector-specific functions and analysis into the usual’s broader impression will observe after the session closes.

Why This Second Issues

When you’ve got spent the previous couple of years watching AI techniques misrepresent your work, your purchasers’ work, or your group’s experience, that is your second to form how that modifications.

The bar for entry is low. You’ll want to overview the specification, check it towards an actual downside you care about, and inform the challenge what you discovered. That suggestions will inform the usual earlier than it turns into finalized.

The session runs for 33 days. After that, the adoption section begins.

Disclosure: I’m the CEO of InLinks and Waikay, which each assist the EntityMap requirements proposal.

Extra Sources:

Featured Picture: optimarc/Shutterstock