The AI industry has a credibility problem, and it is not the fault of the models themselves. When a user asks an AI assistant about a specific company’s product line, the executive team, or the details of a service, the response often arrives with impressive confidence but dubious accuracy. Hallucinated product names, invented executives, and misattributed capabilities have become a routine feature of the generative AI experience. This is not a bug that can be fixed with a larger model or a better prompt. It is a structural failure of the medium. The web was built for human readers who navigate pages, follow links, and interpret prose. AI retrieval systems need something fundamentally different: a structured, verifiable layer of meaning and evidence. EntityMap, a new open standard launched into public consultation today, proposes exactly that.
What Is EntityMap? A Structured Answer to a Structural Problem
EntityMap is an open standard that allows any organization to publish a single, machine-readable file that declares what the organization knows, maps the relationships between its key entities, and links every claim back to its source evidence. The project is currently in a public consultation phase that runs until June 30, 2026, with a formal launch scheduled for July 1. The specification, validation tools, and community forum are all openly available, and the standard is published under a CC BY 4.0 license with no vendor lock-in or proprietary requirements.
How EntityMap Differs From Sitemaps and Schema.org
To understand where EntityMap fits, it helps to look at what already exists. Sitemap.xml tells crawlers which pages exist on a website. Schema.org describes the content that appears on individual pages. Both are essential, but neither was designed to solve the problem of AI retrieval. A sitemap is a directory; schema markup is a label. EntityMap is a knowledge graph.
Consider a healthcare organization that publishes treatment protocols across dozens of pages. With schema.org, you can annotate a single page with a medical condition or a procedure. With EntityMap, you can declare the full scope of institutional knowledge: the core treatment areas, the relationships between them, the peer-reviewed evidence supporting each claim, and the specific page on the site where that evidence lives. An AI system reading that file gets a structured view of the organization’s expertise rather than having to reconstruct it probabilistically from page fragments.
The Anatomy of an EntityMap File
The EntityMap specification is deliberately minimal. A valid file contains three core elements built from roughly twelve required fields.
Entities are the named things an organization covers: products, services, people, concepts, locations, regulations, or areas of expertise. Each entity receives a unique identifier within the file.
Relations map how those entities connect. The standard defines 24 core predicates, including IMPROVES, DEPENDS_ON, MEASURES, LEADS, and GOVERNED_BY. For example, a SaaS company could declare that its product IMPROVES a specific business outcome, or that a particular regulation GOVERNS a service it offers. These relations are what transform a list of entities into a connected knowledge graph.
Evidence chunks are the supporting passages from the website, each linked to its source URL. Every chunk carries attribution metadata: the publisher name, the source page, and a retrieval timestamp. This metadata is designed to survive extraction, aggregation, and storage in vector databases. When an AI system generates a response using that content, the chain of evidence remains intact and traceable.
The specification allows for optional enrichment beyond these core fields: custom predicates, cross-shard resolution, verification status declarations, and changelog tracking are all available for organizations that need deeper granularity.
Why This Matters for Different Stakeholders
The implications of EntityMap cut across multiple roles in the AI and web ecosystem.
For developers building Retrieval Augmented Generation (RAG) systems, cleaner source data means better reasoning chains. A RAG pipeline that ingests an EntityMap file can trace each claim to a specific, authoritative source rather than relying on the incomplete or contradictory fragments that typically result from crawling unstructured content.
For SEO professionals, EntityMap represents a new lever for AI visibility. It works alongside traditional content and link strategies rather than replacing them, but it gives organizations a way to assert semantic control over how their domain knowledge is represented by AI systems.
For publishers, the standard offers a mechanism to preserve attribution as content gets disaggregated across AI platforms. When a passage is retrieved from a vector database, the accompanying metadata ensures the original publisher is credited.
For any organization concerned about AI misrepresentation, EntityMap provides a tool to assert control. A brand can declare its products, differentiate them from competitors, and link to the documentation and case studies that substantiate those claims. An AI system that fetches the EntityMap file gets a definitive, sourced account rather than having to infer differentiation from scattered web content.
Who Is Behind EntityMap and What the Consultation Seeks
The standard originates from within the search and AI community. R.V. Guha, one of the founders of schema.org, has reviewed the project and given it his endorsement. The consultation period is not ceremonial; the project team is actively seeking specific forms of feedback before the standard is finalized.
They need technical implementation feedback from anyone who has tried building an EntityMap for their site or product. What broke? What felt awkward in practice? They need use-case validation to confirm the standard solves real problems and does not miss critical domain-specific requirements. They want critique of the 24 core predicates to determine whether these are the right semantic abstractions for different industries. They are looking for integration ideas from developers building generators, validators, or management dashboards. And they are especially interested in sector-specific applications from healthcare, finance, education, legal, and other verticals that may require specialized profiles.
The Timing and Strategic Significance
The launch of EntityMap arrives at a moment when the limitations of current AI retrieval methods are becoming impossible to ignore. Businesses are discovering that their carefully crafted web content is being reconstructed in unpredictable ways by large language models, often with embarrassing or damaging results. The standard offers a way to address this at the infrastructure level rather than through after-the-fact corrections.
If an organization has spent the last few years watching AI systems misrepresent its work, its products, or its expertise, the consultation period represents an opportunity to shape how that changes. The bar for entry is low: review the specification, test it against a real problem, and provide feedback before June 30, 2026. After that date, the consultation closes and the adoption phase begins.
The specification is available at entitymap.org/spec/v1.0. A validator is live at entitymap.org/validate. The community forum and GitHub repository are at github.com/entitymap. The standard is open, the consultation is genuine, and the opportunity to influence the direction of AI content retrieval is available to anyone willing to participate.
