Skip to content

GEO glossary: AI search terms in plain words

Short definitions of the terms you meet when you look into what AI says about a business. Each is written to be understood without the others.

By the FactHound team. Reviewed October 11, 2026.

Terms

AI assistant
A tool that answers questions in written sentences, such as ChatGPT, Claude, Gemini, Copilot or Perplexity.
AI crawler
A program an AI company runs to read web pages, for training, for search or because a user asked.
AI Overview
The AI-written summary Google shows above some search results.
AI search auditing
Checking what AI tools say about a business and tracing each wrong statement back to its source.
Answer engine optimization (AEO)
Another name for Generative Engine Optimization.
Citation
A link an assistant shows to a page it used for its answer.
Generative Engine Optimization (GEO)
The work of making sure AI assistants can find, read and correctly repeat information about a business.
Grounding
The pages or data an AI answer is based on, as opposed to what the model remembers.
Hallucination
A statement an AI model presents as fact that is not supported by its sources.
JSON-LD
A block of data in a web page that states facts in a form machines can read.
Knowledge Graph
Google’s own database of facts about people, places and businesses.
Large language model (LLM)
The kind of AI model behind these assistants. It writes text one piece at a time from patterns it learned.
LLM brand monitoring
Tracking what AI models say about a business or brand over time.
llms.txt
A proposed plain-text file that points AI tools to a site’s most useful pages.
Local SEO
Making sure a local business’s name, address, phone number and hours are correct and easy to find in search.
Retrieval-augmented generation (RAG)
An assistant fetching pages first and then writing its answer from them.
robots.txt
The file on a website that tells crawlers what they may fetch.
schema.org
The shared vocabulary used in structured data.
Structured data
Facts written in a fixed format, such as JSON-LD, so that software doesn’t have to guess.
Training data
The text a model learned from, gathered up to a cutoff date.
Verified fact
In FactHound, a fact about your business that you have confirmed is true.

Where FactHound helps, and where it doesn’t

What FactHound does

  • FactHound keeps one list of verified facts per website, publishes it in the formats above and checks AI answers against it.

What it doesn’t do

  • FactHound covers the accuracy side of GEO. It doesn’t measure how often you are mentioned or where you rank.

Start with the free check

No signup. Enter your website to see whether AI crawlers can read it and which questions it can’t answer.

Run the free check