The AI search glossary

66 AI terms and definitions for anyone working on search in the age of AI assistants: the AI concepts, the SEO terminology, and the GEO glossary in between, each in plain English with a link to go deeper.

Updated 26 September 2026

AI search brought its own vocabulary, and much of it overlaps with the SEO glossary people already know. GEO, AEO, LLMO and AI SEO describe largely the same work; RAG and grounding explain why some AI answers can cite you and others cannot; GPTBot, OAI-SearchBot and Google-Extended decide who can read your site in the first place.

It is not a general glossary of artificial intelligence. This AI glossary of terms keeps to the words that matter when AI assistants answer your customers’ questions: how they find pages, how they cite them, and how you measure it.

Search for a term, filter by topic, or jump by letter. Every entry has its own link, so you can send a colleague straight to what RAG is or what query fan-out means.

Terms by topic

How this glossary of AI search terms is split.

Terms by topic
AI and LLM concepts	12%
AI search surfaces	10%
AI visibility and GEO	12%
Crawlers and access	12%
Measurement	6%
SEO and content	14%

66 terms

A

AEO (answer engine optimization)

AI visibility and GEO

Also: answer engine optimization, answer engine optimisation

Optimising content so answer engines and AI assistants can lift it as a direct answer. It overlaps heavily with GEO; AEO grew out of voice search and featured snippets, GEO out of generative AI.

Answer Engine Optimization →GEO (generative engine optimization)Answer engine

Agentic commerce

AI search surfaces

Also: Agentic Commerce Protocol, ACP, ChatGPT shopping, instant checkout

Shopping carried out through an AI assistant: finding, comparing and sometimes buying products in the chat. OpenAI's open Agentic Commerce Protocol lets merchants share product feeds with ChatGPT.

ChatGPT Shopping →AI agent Structured data

AI agent

AI and LLM concepts

Also: agentic ai, agentic search, agent

An AI system that takes actions on a user's behalf, such as browsing sites, filling forms or buying something, rather than only answering. Operators publish user agents for some of these fetchers.

Agentic commerce User agent

AI Assistant channel (GA4)

Measurement

Also: ga4 ai assistant, ai-assistant medium

A channel in Google Analytics 4's default channel group since 13 May 2026, for visits from assistants such as ChatGPT, Gemini, DeepSeek, Copilot and Grok. It excludes Google's AI Overviews and AI Mode.

How to measure AI traffic in GA4 →AI referral traffic UTM parameters

AI crawler

Crawlers and access

Also: ai bot, ai web crawler

A crawler run by an AI company. There are three kinds: training crawlers that collect data for models, search crawlers that build an AI assistant's index, and user-triggered fetchers that read a page when someone asks about it.

AI Traffic Analytics →GPTBot ClaudeBot OAI-SearchBot

AI Mode

AI search surfaces

Also: Google AI Mode

Google's conversational AI search, reached through a tab or button in Google Search. It launched in Labs in March 2025 and opened to everyone in the US in May 2025, and uses query fan-out to research a question.

Google AI Mode →Query fan-out AI Overview

AI Overview

AI search surfaces

Also: AI Overviews, Google AI Overview, SGE, Search Generative Experience

The AI-generated summary Google shows at the top of some search results, with links to the pages it drew on. It grew out of the Search Generative Experience and launched widely in the US in May 2024.

Why it matters: AI Overview citations lean on pages that already rank, but not only on the top ten. Clicks from them are counted as ordinary organic search.

Google AI Overviews →AI Mode Zero-click search Featured snippet

AI referral traffic

Measurement

Also: llm traffic, ai traffic, chatgpt referral traffic

Visits that arrive from an AI assistant, such as a click on a link in a ChatGPT or Perplexity answer. Analytics tools see them as referrals from sites like chatgpt.com.

Why it matters: It is always undercounted: some apps send no referrer, and clicks from Google's AI features count as organic search.

How to measure AI traffic in GA4 →AI Assistant channel (GA4)Referrer

AI SEO

AI visibility and GEO

A loose term with two meanings: using AI tools to do SEO work, and doing SEO so that AI systems find and cite you. Most people searching for it mean one or the other, rarely both.

AI SEO →GEO (generative engine optimization)AEO (answer engine optimization)

AI visibility

AI visibility and GEO

Also: ai search visibility, llm visibility, visibility score

How often AI assistants mention your brand for the questions your buyers ask, usually expressed as the share of tracked answers that name you.

Why it matters: Because answers vary run to run, AI visibility is only meaningful measured over many prompts and repeated runs.

AI Visibility Tracking →Share of voice Prompt tracking Non-determinism

Answer engine

AI search surfaces

Also: AI answer engine, AI search engine, generative search

A search product that replies with a written answer instead of a list of links, such as Perplexity, ChatGPT search or Google's AI Mode.

Answer Engine Optimization →AEO (answer engine optimization)Zero-click search

B

Backlink

SEO and content

Also: inbound link, link building

A link from another site to yours. Links remain a signal of trust for search engines and a route by which AI systems discover pages.

Brand mention Entity

Brand mention

AI visibility and GEO

Also: mention, unlinked mention

Your brand being named in an AI answer or on a web page, with or without a link. In AI search, being named is often worth more than being linked, because many answers name brands without citing anyone.

Citation (AI citation)AI visibility

C

Canonical tag

SEO and content

Also: canonical url, rel canonical

A tag that tells search engines which URL is the main version of a page when the same content is available at more than one address.

Site Audit →XML sitemap

ChatGPT search

AI search surfaces

Also: SearchGPT

ChatGPT's web search mode. When it searches, answers include inline citations and a Sources panel. OpenAI's OAI-SearchBot crawler is what surfaces websites in it.

ChatGPT SEO →OAI-SearchBot Citation (AI citation)

Citation (AI citation)

AI visibility and GEO

Also: ai citation, llm citation, source

A link an AI answer shows as its source for a statement. Engines that search the web, such as Perplexity, ChatGPT search and Google's AI features, cite; a model answering from memory cannot.

How AI assistants decide which sources to cite →Grounding Source gap Brand mention

ClaudeBot

Crawlers and access

Anthropic's crawler for content that could contribute to training its Claude models. It honours robots.txt and Crawl-delay. Anthropic also runs Claude-SearchBot for search and Claude-User for user requests.

AI Traffic Analytics →GPTBot Crawl-to-refer ratio

Confidence interval

Measurement

Also: error bars, margin of error, wilson interval

A range that shows how precise an estimate is. A share of voice of 20% measured from a small sample might really be anywhere from 12% to 30%; two brands whose ranges overlap cannot honestly be ranked.

Competitor Tracking →Share of voice Non-determinism

Content brief

SEO and content

A plan for a page before it is written: the question it answers, what it must cover, the sources to use and how it will be judged.

Content Briefs →Search intent Source gap

Context window

AI and LLM concepts

The amount of text, counted in tokens, a model can take into account at once: the question, any retrieved pages and its own answer so far.

Why it matters: Retrieved pages compete for limited space. A page that answers the question early and plainly is easier to use than one that buries it.

Token RAG (retrieval-augmented generation)

Core vs Carousel

AI visibility and GEO

Also: core position, carousel position

CoreCited's name for the difference between a brand AI names in almost every answer to a question (Core) and one that rotates in and out (Carousel). It is judged over several weekly runs, not one.

Core vs Carousel →Non-determinism AI visibility

Core Web Vitals

SEO and content

Also: CWV, LCP, INP, CLS

Google's page experience metrics for loading (Largest Contentful Paint), responsiveness (Interaction to Next Paint) and visual stability (Cumulative Layout Shift).

Site Audit →Server-side rendering (SSR)

Crawl-to-refer ratio

Crawlers and access

How many pages an AI platform crawls for every visitor it sends back. A high ratio means a platform reads a lot and sends little traffic.

AI Traffic Analytics →AI crawler AI referral traffic

E

E-E-A-T

SEO and content

Also: EEAT, E-A-T, experience expertise authoritativeness trustworthiness

Experience, expertise, authoritativeness and trustworthiness: the qualities Google's search quality rater guidelines ask raters to look for. Google added the first E, experience, in December 2022, and says trust matters most.

E-E-A-T when the reader is a machine →YMYL Topical authority

Embedding

AI and LLM concepts

Also: vector, vector embedding

A list of numbers that represents the meaning of a piece of text, so that texts with similar meanings have similar numbers. Embeddings power semantic search and much of the retrieval step in RAG.

Semantic search RAG (retrieval-augmented generation)

Entity

SEO and content

Also: knowledge graph, named entity

A distinct thing a search engine or model recognises, such as a company, person, product or place, with facts attached to it. Google's Knowledge Graph is a database of entities.

Why it matters: When your brand is a clear entity with consistent facts across the web, AI systems are less likely to confuse it or get it wrong.

Structured data Hallucination

F

Featured snippet

AI search surfaces

A short extract from one web page that Google shows at the top of some results, quoting the page directly and linking to it. It predates AI Overviews, which write a new answer from several sources.

Google AI Overviews →AI Overview SERP

G

GEO (generative engine optimization)

AI visibility and GEO

Also: generative engine optimization, generative engine optimisation

The practice of making a brand and its content more likely to appear, and be cited, in AI-generated answers. The term comes from a 2023 research paper that tested which content changes raised visibility in generative search.

Generative Engine Optimization →AEO (answer engine optimization)AI SEO LLMO (LLM optimization)

Google-Extended

Crawlers and access

A robots.txt token, not a separate crawler, that controls whether Google may use your content to train Gemini models and ground Gemini apps. Google says it does not affect inclusion in Google Search, including AI Overviews and AI Mode.

Robots.txt Generator & Tester →robots.txt AI Overview

GPTBot

Crawlers and access

OpenAI's training crawler. Blocking it in robots.txt keeps your future content out of OpenAI's model training; it does not remove you from ChatGPT search, which uses OAI-SearchBot.

Should you block GPTBot? →OAI-SearchBot robots.txt Training data

Grounding

AI and LLM concepts

Also: grounded answer, grounding ai, web grounding

Tying an AI answer to specific sources, such as search results or documents, so its claims can be checked against them. A grounded answer usually shows citations; an ungrounded one comes from the model's memory alone.

Why it matters: Only grounded answers can cite your page. Engines that answer from memory can name your brand but have no link to give.

Citation Analysis →RAG (retrieval-augmented generation)Citation (AI citation)Hallucination

H

Hallucination

AI and LLM concepts

Also: ai hallucination, confabulation

When an AI model states something false with confidence, such as a wrong price, a feature a product does not have, or a source that does not exist.

Why it matters: Hallucinations about your brand are a reason to track what AI says about you, not only whether it mentions you. Clear, consistent facts across the web give models less room to guess.

Grounding Sentiment

Headless CMS

SEO and content

A content management system that stores content and serves it through an API, leaving a separate front end to build the pages. Contentful, Sanity, Strapi and Storyblok are headless.

Why it matters: SEO settings such as titles, sitemaps and rendering then live in the front end's code.

CMS Publishing →Server-side rendering (SSR)

I

Impressions

Measurement

How many times a page appeared in search results, whether or not anyone clicked. Search Console's Generative AI report counts impressions in Google's AI features.

Zero-click search AI Overview

Internal linking

SEO and content

Links between pages on the same site. They help crawlers find pages and show how topics connect.

Topical authority

J

JSON-LD

SEO and content

The format Google recommends for structured data: a block of JSON inside a script tag, kept separate from the visible HTML.

Schema Markup Generator →Structured data

L

Large language model (LLM)

AI and LLM concepts

Also: LLM, language model, foundation model

An AI model trained on a very large amount of text to predict the next word, which lets it answer questions, write and summarise. GPT (behind ChatGPT), Claude, Gemini and Llama are large language models.

Why it matters: An LLM can only mention your brand from what it learned in training or what it retrieves while answering, so both routes matter.

Training data RAG (retrieval-augmented generation)Token

LLMO (LLM optimization)

AI visibility and GEO

Also: LLM SEO, LLM optimization, LLM optimisation

Another name for optimising for large language models, used interchangeably with GEO. The work is the same: be findable, quotable and talked about in the sources models use.

GEO vs SEO vs AEO →GEO (generative engine optimization)AI SEO

llms.txt

Crawlers and access

Also: llms txt, llms-full.txt

A proposed Markdown file at yoursite.com/llms.txt that summarises a site and links to its most useful pages for AI systems. Jeremy Howard proposed it in September 2024. Google says its Search and AI features do not use it.

Does llms.txt actually do anything? →robots.txt XML sitemap

N

Non-determinism

AI and LLM concepts

Also: answer variability, ai inconsistency

The same prompt to the same AI model can produce a different answer each time. For brand recommendations the variation is large: one study found less than a 1 in 100 chance of the same list of brands twice.

Why it matters: This is why one check proves little. AI visibility has to be measured across repeated runs.

How AI assistants pick the products they recommend →Core vs Carousel Confidence interval

O

OAI-SearchBot

Crawlers and access

OpenAI's crawler for ChatGPT search. OpenAI says sites that opt out of it are not shown in ChatGPT search answers, though they can still appear as navigational links.

AI Crawler Checker →GPTBot ChatGPT search

P

Perplexity

AI search surfaces

An AI answer engine that searches the web for every answer and shows numbered citations to its sources. Its crawler is PerplexityBot.

Citation Analysis →Answer engine Citation (AI citation)

Prompt

AI and LLM concepts

Also: tracked prompt, buyer prompt

The question or instruction given to an AI model. In AI visibility work, a prompt is a question your buyers would ask, such as 'best CRM for a small agency', tracked over time.

AI Visibility Tracking →Prompt tracking AI visibility

Prompt tracking

AI visibility and GEO

Also: ai rank tracking, llm tracking

Running a fixed set of buyer questions through AI assistants on a schedule and recording which brands and sources each answer includes.

AI Visibility Tracking →Prompt AI visibility

Q

Query fan-out

AI search surfaces

Also: fan-out

The technique Google describes for AI Mode: breaking a question into subtopics and running many searches at once, then combining the results into one answer.

Why it matters: A page can be cited for a sub-question it answers well even if it would never rank for the original query. Covering the related questions around a topic widens your chances.

Google AI Mode →AI Mode Topical authority

R

RAG (retrieval-augmented generation)

AI and LLM concepts

Also: retrieval augmented generation

RAG is a technique where an AI system first retrieves relevant documents, for example by searching the web or a company's files, and then writes its answer using them. The term comes from a 2020 research paper by Lewis and colleagues at Facebook AI.

Why it matters: RAG is why AI search can cite pages published last week: the answer is built from what was retrieved, not only from training. It is also why your pages need to be crawlable and easy to quote.

How AI assistants decide which sources to cite →Grounding Citation (AI citation)Semantic search

Referrer

Measurement

Also: referer, referral

The address of the page a visitor came from, passed by the browser. When an app does not pass one, analytics records the visit as direct.

AI referral traffic UTM parameters

robots.txt

Crawlers and access

Also: robots file, robots exclusion protocol

A text file at the root of a site that tells crawlers which paths they may fetch, written as groups of rules per user agent. Well-behaved crawlers follow it; it is a request, not a lock.

Why it matters: It is where you choose which AI crawlers can read your site, and a mistake here can remove you from AI search entirely.

Robots.txt Generator & Tester →User agent GPTBot llms.txt

S

Search intent

SEO and content

Also: user intent, keyword intent

What a person actually wants from a search: to learn something, to find a specific site, to compare options, or to buy.

Prompt Content brief

Semantic search

AI and LLM concepts

Also: vector search

Search that matches on meaning rather than exact words, so a page about 'cutting cloud costs' can be found for 'reduce AWS bill'.

Why it matters: It rewards pages that cover a topic thoroughly in natural language over pages that repeat one phrase.

Topical authority for AI answers →Embedding Query fan-out

Sentiment

AI visibility and GEO

Also: brand sentiment

Whether an AI answer describes your brand positively, neutrally or negatively. A mention that recommends a competitor over you counts differently from one that recommends you.

Brand mention Hallucination

SERP

AI search surfaces

Also: search engine results page

The page a search engine shows for a query: the links, ads, AI Overview and any other features.

Featured snippet AI Overview

Server-side rendering (SSR)

Crawlers and access

Also: SSR, pre-rendering, static generation

Building a page's HTML on the server, so the content is in the response crawlers receive. The alternative, client-side rendering, builds the page with JavaScript in the browser.

Why it matters: Many AI crawlers do not run JavaScript, so content that only appears after scripts run can be invisible to them.

CMS Publishing →Headless CMS Web crawler

Source gap

AI visibility and GEO

Also: citation gap

The sites AI answers cite when they recommend your competitors but have never cited when naming you. It turns a visibility score into a list of pages to work on.

Citation Analysis →Citation (AI citation)Share of voice

Structured data

SEO and content

Also: schema markup, schema.org

Code added to a page, usually in the schema.org vocabulary, that labels what the page contains: an article, a product and its price, an organisation, a set of FAQs.

Why it matters: It removes guesswork for machines reading the page, including product details that AI shopping answers use.

Schema Markup Generator →JSON-LD Entity

T

Title tag and meta description

SEO and content

Also: meta title, meta description, title tag

The page title and short summary in a page's HTML head. Search engines often show them in results, and they tell any reader, human or machine, what the page is about.

AI SEO Checker →Structured data

Token

AI and LLM concepts

The unit a language model reads and writes, usually a word or part of a word. Model limits and API prices are counted in tokens.

Context window Large language model (LLM)

Topical authority

SEO and content

The idea that a site covering a subject thoroughly, across many connected pages, is treated as a stronger source on it than a site with one page.

Topical authority for AI answers →Query fan-out Internal linking

Training data

AI and LLM concepts

Also: training cutoff, knowledge cutoff, pretraining

The text a model learned from before release. Anything published after the training cutoff is unknown to the model unless it retrieves it while answering.

Why it matters: Being widely discussed before a model's cutoff helps it recall your brand without searching. Training crawlers such as GPTBot and ClaudeBot collect this data.

GPTBot ClaudeBot RAG (retrieval-augmented generation)

U

User agent

Crawlers and access

Also: UA, user-agent string

The name a browser or crawler sends with each request, such as GPTBot or Googlebot. robots.txt rules are written per user agent, and server logs record it.

robots.txt Web crawler

UTM parameters

Measurement

Also: utm, utm_source

Tags added to a link, such as utm_source=chatgpt.com, that tell analytics where a visit came from even when the referrer is missing. OpenAI says ChatGPT adds utm_source=chatgpt.com to links automatically.

How to measure AI traffic in GA4 →Referrer AI Assistant channel (GA4)

W

Web crawler

Crawlers and access

Also: crawler, spider, bot

A program that fetches web pages automatically, for a search index, an AI system or an archive. Crawlers identify themselves with a user agent.

AI Traffic Analytics →AI crawler User agent robots.txt

X

XML sitemap

Crawlers and access

Also: sitemap, sitemap.xml

A file listing the URLs on a site you want crawlers to find, often with the date each was last changed.

robots.txt llms.txt

Y

YMYL

SEO and content

Also: your money or your life

'Your money or your life': topics such as health, finance and safety where Google's guidelines hold content to a higher standard, because bad information could cause real harm.

E-E-A-T

Z

Zero-click search

AI search surfaces

Also: zero click

A search that ends without the user clicking through to any website, because the answer was on the results page or in an AI summary.

What zero-click search means for your traffic →AI Overview Featured snippet

Keep reading

GuideGEO vs SEO vs AEO What actually differs, and what to do first.GuideAI SEO Using AI for SEO, and doing SEO for AI.GuideGoogle AI Overviews How a page ends up inside the box above the results.ArticleDoes llms.txt actually do anything? The sceptical companion to the generator.Free toolRobots.txt Generator & Tester Build a robots.txt and test any URL against it.FeatureAI Visibility Tracking Weekly tracking across six AI surfaces.