# Becoming an entity — how AI learns who you are

*2026-06-07 · Discoverability & Structure*

> The unit of the AI web is the entity, not the page. Here's how to get recognised in the knowledge graph so assistants can reason about who you are with confidence.

## Pages versus entities

Classic SEO thinks in **pages**: this URL ranks for that query. AI systems increasingly
think in **entities**: a person, a company, a product, a concept — a defined node with
attributes and relationships. When an assistant says "datavision.ie is a consultancy
founded by Paul Masterson specialising in X," it isn't reading a page; it's reciting an
**entity** it has learned and trusts.

If the graph doesn't know you as an entity, the best you can hope for is to be
approximated from scattered pages. Becoming an entity is criterion **D4** in the
[Discoverability & Structure pillar](/framework#discoverability), and it's how you go
from "mentioned" to "understood."

## What a knowledge graph is

A knowledge graph is a structured map of entities and the relationships between them.
There's Google's proprietary one behind its Knowledge Panels, and there's the open,
public one — chiefly **Wikidata** — that many AI systems draw on. The goal is the same:
to have a stable, canonical node that says *this is who they are*, which other systems
can reference.

## How to build entity presence

Five reinforcing moves:

1. **Consistent identity everywhere.** One exact business name, one canonical spelling, consistent address and details across every profile and directory. Inconsistency fragments the entity.
2. **`sameAs` links.** In your Organization and Person schema, list `sameAs` URLs pointing to your authoritative profiles — LinkedIn, Crunchbase, professional bodies, social accounts. These are the threads that stitch your scattered presence into one node. (This builds directly on a [verifiable author identity](/guides/verifiable-author-identity).)
3. **Structured data.** Strong `Organization` and `Person` schema — the machine-readable description of the entity itself. See [schema markup](/guides/what-is-llms-txt) and the structured-data criterion.
4. **Authoritative-graph presence.** Where genuinely notable and within the rules, a well-sourced Wikidata entry; otherwise, presence in the registries and directories relevant to your field.
5. **Corroboration.** Entities are believed when multiple independent sources agree. Citations, mentions, and profiles that all say the same consistent thing strengthen the node.

## Checking you're a recognised entity

Test it the way an assistant would: ask the major engines "who is *your-brand*?" and "tell
me about *your-name*." If the answers are confident, accurate, and consistent, you have an
entity. If they're vague, hedged, or wrong, you have pages — and a gap to close. (This is
where [AI-representation observability](/guides/ai-representation-observability) and entity
work meet.)

## What to do this week

1. Audit your business name, details, and profile links for consistency across the web — fix the mismatches.
2. Add or complete `sameAs` arrays in your Organization and Person schema.
3. Ask the major assistants who you are and record how accurately they answer.
4. If you're genuinely notable, scope a properly-sourced Wikidata entry; if not, target the key directories in your field.