What is an entity graph?

An entity graph is a machine-readable map of who and what you are and how you connect to other known entities — your company, its founder, its category, and the external profiles that confirm all of it. AI engines use it to resolve your identity before deciding whether to cite you.

The entity graph: how AI decides who to trust

By Mukund Swaroop · 6 July 2026

TL;DR

AI engines resolve entities before they cite them: they match your brand to a known thing using structured data and consistent links across the web. A clean, bidirectional entity graph — a Person and an Organization joined by stable identifiers and sameAs links — lets an engine confirm who you are in one hop, which is the precondition for being recommended.

This essay is in progress. The outline and argument are set; the full draft is being written.

Why recognition comes before citation

[[DRAFT]] — An engine won’t stake its credibility on a company it can’t identify. Entity resolution is the gate; citation is downstream of it.

What entity resolution actually is

[[DRAFT]] — Matching a string (“your brand”) to a node (a known thing). How models disambiguate using structured data and corroborating signals.

The role of structured data and sameAs

[[DRAFT]] — schema.org Person and Organization nodes, stable @id anchors, and sameAs links to trusted profiles as the chain of trust an engine can traverse.

Bidirectional links: the founder ↔ company loop

[[DRAFT]] — Person.worksFor → Organization and Organization.founder → Person. Why the reciprocal edge lets one query surface both, and how to mirror the same identifier across two domains.

How to build your own entity graph

[[DRAFT]] — A practical checklist: consistent name string, one canonical description, shared @id across properties, and only real sameAs targets (a broken one is worse than none). Wikidata and ORCID as high-leverage anchors.