Google Knowledge Graph Explorer

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://site.example/#entity",
  "url": "https://site.example",
  "name": "Anthropic",
  "description": "Anthropic is an American AI startup company...",
  "sameAs": [
    "https://en.wikipedia.org/wiki/Anthropic",
    "https://www.google.com/search?kgmid=/g/11h_2zs0j3"
  ]
}

How to read the result

• In the Knowledge Graph (green) — at least one entity matched your query. The top match is highlighted; up to three matches are shown in case the name is ambiguous.
• Not found (red) — nothing matched, so the entity isn't in the Knowledge Graph yet. You get the not-found playbook instead of an entity card.
• Type chips — the schema.org types Google assigns the entity (Organization, Person, Place…). Thing is the catch-all and is dropped from the generated JSON-LD unless it's all Google has.
• KG ID (mid) — the stable machine identifier, e.g. /m/0d6lp. The "see the knowledge panel" link opens google.com/search?kgmid=… for that exact entity.
• Match confidence — how well the entity matched your query text, not an authority or "entity strength" score. It's absent for entities returned by the newer API, and Google is phasing it out.

How it works

Each lookup calls a small server endpoint (/api/kg-lookup) that queries the official Google Knowledge Graph Search API with your name and any type filter. The raw response is normalised down to the fields shown — name, mid, types, description, the Wikipedia summary, image, and the match score — and cached for up to 24 hours (misses are cached too, so repeat "not found" checks are cheap).

The JSON-LD is built in your browser from that normalised entity: it picks the most specific schema.org type Google reported, keeps the name and description, and assembles a sameAs array from the Wikipedia URL and the Google KG ID. Add your own site URL and it sets url and an @id so you claim the entity's home page as yours. Nothing you look up is stored; descriptions shown come from Wikipedia (CC BY-SA).

Features

• Found / not-found verdict against live Google Knowledge Graph data, top three matches.
• Optional entity-type filter (Person, Organization, Corporation, Place, Product, and more) to disambiguate common names.
• Entity card with description, Wikipedia summary, type chips, image, and the Knowledge Graph ID plus a direct link to its knowledge panel.
• Copy-paste schema.org JSON-LD generated for the entity, with an optional site URL to claim url/@id.
• A concrete not-found playbook for getting an entity recognised.
• Side-by-side comparison of two entities and shareable links for any lookup or comparison.

Limitations

This checks Google's Knowledge Graph Search API — not the live knowledge panel, which can differ and isn't publicly queryable. A match means Google recognises the entity string, not that a knowledge panel renders for it. The free comparison covers two entities at a time. The match confidence is a query-relevance number, not authority, so don't read it as entity strength. And Google does not publish a guaranteed Knowledge Graph update timeline, so a recent change may not appear in this API and this tool cannot predict when it will.

Frequently asked questions