Glossary
The AI search glossary
This is a plain-English dictionary for AI search. If you keep running into terms like GEO, AEO, AI visibility, or AI citation and you are not totally sure what they mean, you are in the right place. Every definition here is written the way you would explain it to a coworker, not the way a tool vendor would pitch it.
Here is the quick map. GEO (Generative Engine Optimization) is the work of getting your content quoted inside AI answers, not just ranked in a list of links. AEO (Answer Engine Optimization) is the close cousin focused on being the answer a tool gives back. AI visibility is the big-picture question of whether engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews mention you at all. An AI citation is the actual moment one of them names your site as a source. The rest of the words below (share of voice, fan-out queries, grounding, RAG, llms.txt, and so on) are the gears underneath those ideas.
So why a glossary? Because the language is new and it moves fast, and half the confusion in this space is just people using the same word to mean three different things. We tried to fix that. Pick a term, get a clear answer, and move on. And if you only read one entry, make it GEO.
Agentic commerce
Agentic commerce is when AI shopping agents act on behalf of the shopper. You hand the agent a goal (best X under $Y by Friday), and it researches, compares, and increasingly buys across merchants. Brands win by having clean, well-reviewed, structured product data that matches the shopper's intent and price, because that is what the agent reads before it picks.
AI agent
An AI agent is an AI system that doesn't just answer questions, it takes actions toward a goal. It can search the web, use tools, compare options, and finish multi-step tasks with little human input along the way. A chatbot tells you the three best running shoes; an agent finds them, checks the details, and can put a pair in the cart.
AI citation
An AI citation is when a generative AI engine (such as ChatGPT, Perplexity, Google AI Overviews, Claude, or Gemini) references your content as a source inside the answer it gives a user. That reference can be a clickable link, a numbered footnote, or just your brand named in the text. In AI search and GEO, citations are the new visibility: getting cited means the AI is using and crediting your content, even when nobody clicks through.
AI crawler
An AI crawler is a bot that fetches web content for AI systems, either to train a model, to build a search index, or to pull a live page while it writes an answer. The big names include GPTBot and OAI-SearchBot (OpenAI), PerplexityBot, ClaudeBot (Anthropic), and Google-Extended. Block the search-flavored ones and you can quietly vanish from that engine's answers.
AI hallucination
An AI hallucination is when an AI model states something false or made up as if it were fact: a fake statistic, a citation that does not exist, a wrong price, or a feature your brand never shipped. It happens because models predict plausible text, not verified truth. For brands that is a real risk, because a confident wrong answer about you can spread fast.
AI Mode
AI Mode is Google's Gemini-powered conversational search experience built right into Google Search. You ask long, messy, multi-step questions, follow up in context, and it hands back one synthesized answer plus supporting links. Under the hood it splits your question into many sub-searches, which is what separates it from an AI Overview, the summary box that sits on top of a normal results page.
AI visibility
AI visibility is the measure of how often and how prominently your brand, content, or website shows up inside the answers generated by AI engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews. Think of it as a mention rate, not a ranking: the question is how frequently you get named or cited across many different prompts. Where SEO visibility is about ranking on a results page, AI visibility is about being part of the answer itself.
AI Visibility Score
An AI Visibility Score is a single metric, usually shown as a percentage, that measures how often and how prominently your brand appears in answers from AI engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews across a set of test prompts. At its simplest it's the share of tracked prompts where you get mentioned. More advanced versions weight that by where you appear in the answer, whether you're cited with a link, and how the AI talks about you.
Answer engine
An answer engine is a search tool that uses AI and large language models to read your question in plain language and hand back one direct, synthesized answer, instead of a list of blue links you have to click through. Examples include ChatGPT, Perplexity, Google AI Overviews, and Microsoft Copilot. It still pulls from web sources and often cites them, but the goal is to answer you on the spot.
Answer Engine Optimization (AEO)
Answer Engine Optimization (AEO) is the practice of writing and structuring content so AI answer engines (like ChatGPT, Perplexity, and Google AI Overviews) can understand it, trust it, and quote it as the direct answer to a question. The goal is being cited inside the answer, not just ranking a link. Unlike classic SEO, which chases clicks, AEO chases citations.
Brand mention vs citation
A brand mention is when an AI engine says your brand name inside an answer, with no link attached. A citation is when the AI points to your specific page as the source of a claim, usually with a link or footnote. The short version: a mention is the AI talking about you, a citation is the AI crediting you. Both help your AI visibility, but they're not the same thing and they're earned in different ways.
Citation rate
Citation rate is the share of AI-generated answers, across a set of tracked questions, where your brand or page shows up as a cited source. The math is simple: answers that cite you divided by total answers checked, times 100. So if you run 100 prompts through ChatGPT and your site is cited in 22 of the answers, your citation rate is 22%. It's the clearest single number for how often AI engines are actually using and crediting your content.
Conversational AI
Conversational AI is technology that lets people interact with software using everyday natural language, back and forth, the way you'd talk to a person instead of typing keywords. Chatbots, voice assistants like Siri and Alexa, and AI search tools like ChatGPT, Perplexity, and Gemini are all conversational AI. It moved search away from typing three keywords and toward asking a full question, which is exactly why answer-first content wins now.
DeepSeek
DeepSeek is a Chinese AI lab whose open-weight models became widely used across 2025 and 2026 for hitting strong performance at a famously low cost. Because the weights are open, those models show up inside chatbots, apps, and search tools all over the place, and some AI visibility platforms now track DeepSeek as one more engine that can mention or cite a brand. AI Citation Monitor does not track DeepSeek yet. We track ChatGPT, Perplexity, Gemini, Google AI Overviews, and Microsoft Copilot today.
E-E-A-T
E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trustworthiness, the quality framework Google describes in its Search Quality Rater Guidelines to judge how trustworthy a source is. AI engines lean on similar signals when they choose which pages to quote. Strong E-E-A-T (named expert authors, real credentials, citations, accurate facts) makes your page a safer source for an AI to put its name behind.
Fan-out query
A fan-out query is when an AI engine takes your single question and quietly splits it into several related sub-searches, then gathers sources across all of them before writing one answer. It is why you can get cited for a question you never directly targeted: you covered an adjacent sub-topic well, and the fan-out found you. Think of it as one prompt fanning out into a small fleet of searches running behind the scenes.
Featured snippet
A featured snippet is the short answer box Google pulls from a single ranking page and shows near the top of search results. For years it was the prize spot, the famous "position zero." In 2026 the AI Overview usually sits above it and blends several sources into one answer, so the featured snippet is no longer the first thing many searchers actually see.
Generative AI
Generative AI is artificial intelligence that creates new content (text, images, code, audio) from a prompt, rather than just classifying or predicting from existing data. The large language models behind ChatGPT, Gemini, Claude, and Perplexity are generative AI, and they are what turned search from a list of links into a written answer.
Generative Engine Optimization (GEO)
Generative Engine Optimization (GEO) is the practice of structuring and writing content so AI engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews pull it into their answers and cite it as a source. Where SEO aims to get you clicked from a list of blue links, GEO aims to get you quoted inside the answer itself. The term comes from a 2023 Princeton-led research paper that first measured what makes content show up in AI-generated responses.
Google AI Overviews
Google AI Overviews are AI-generated summaries that show up at the top of Google Search results, above the regular blue links. Powered by Google's Gemini models, they answer your question right on the page by pulling and stitching together information from across the web, with a few source links attached. By 2026 they appear on roughly half of all searches and reach over 2 billion people a month, which makes them one of the biggest forces in how content gets found.
Grok
Grok is xAI's conversational AI assistant, built into X (formerly Twitter) and available as a standalone app and on the web. For brand visibility it matters because studies have clocked Grok with one of the highest source citation rates of any AI engine, meaning it links out to real sources more often than most. AI Citation Monitor tracks ChatGPT, Perplexity, Gemini, and Google AI Overviews today, not Grok.
Grounding
Grounding is when an AI model builds its answer from specific external sources (live web results, a document set, a database) instead of relying only on its training memory, and it usually cites what it used. Grounding is what makes citations possible. It's the door your brand gets cited through.
Knowledge graph
A knowledge graph is a structured network of entities (people, places, brands, products) and the relationships between them, stored as machine-readable facts instead of paragraphs. Google Knowledge Graph and Wikidata are the two big public ones. Search engines and AI use them as a fact layer to disambiguate and verify a brand before describing or citing it, so a clean, well-connected entity makes AI more confident naming you.
Large language model (LLM)
A large language model (LLM) is an AI system trained on massive amounts of text to predict and generate language, which lets it answer questions, summarize, and write. LLMs like GPT, Gemini, Claude, and Llama power the AI search engines and chatbots people now use instead of typing into a search bar. For brands, the part that matters is simple: when an LLM answers a question, it decides which sources to quote, and you want to be one of them.
LLM SEO
LLM SEO (Large Language Model SEO) is the practice of optimizing your content so AI models like ChatGPT, Gemini, Claude, and Perplexity understand it, trust it, and cite it in their answers. Instead of chasing a top spot in a list of blue links, you're trying to be the source the AI quotes when it replies to a question. It overlaps heavily with GEO and AEO, and most people use the three terms to mean roughly the same thing.
LLMO (large language model optimization)
LLMO (large language model optimization) is the practice of shaping your content so large language models like ChatGPT, Claude, and Gemini can understand it, trust it, and cite it in their answers. It is basically another name for GEO (generative engine optimization) and AI SEO. The labels overlap heavily and point at the same underlying job: getting picked up and quoted by AI.
llms.txt
llms.txt is a plain Markdown file you place at yoursite.com/llms.txt that hands AI models a curated map of your most important pages. Jeremy Howard of Answer.AI proposed it in September 2024. It is an emerging convention, not a confirmed ranking factor, and no major AI provider has confirmed that their crawlers actually read it.
Microsoft Copilot
Microsoft Copilot is Microsoft's AI assistant, built on large language models and woven through Windows, Edge, Bing, and Microsoft 365 apps. When it answers a question about the live web, it grounds the reply in Bing search: it turns your prompt into a short Bing query, reads those results, and shows a Sources panel with the pages it leaned on. So if you want Copilot to cite you, strong Bing visibility is the lever you pull.
Position zero
Position zero is the answer slot that sits above the number one organic result. Historically it meant the featured snippet, that boxed answer Google pulled to the top of the page. In 2026 it usually means the AI Overview, which sits above everything else: snippets, organic links, and ads. Winning position zero now means being one of the sources the AI summarizes, not just ranking first.
Prompt tracking
Prompt tracking is the practice of running a fixed list of prompts through AI engines like ChatGPT, Perplexity, and Gemini on a repeating schedule, then recording whether your brand gets mentioned, cited, or recommended in the answers. It's the AI-search version of keyword rank tracking: instead of watching where a page ranks, you watch how your brand shows up across many real-world questions. The prompts stay the same week to week so you can spot trends, catch drops, and benchmark against competitors.
Retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is a technique where a language model fetches relevant external documents at query time and grounds its answer on them, instead of relying only on what it memorized during training. It is the core mechanism behind AI search: ChatGPT Search, Perplexity, Gemini grounding, and Google AI Overviews all retrieve live web pages and then generate an answer from them. For anyone doing GEO, this is the whole game. If your page lands in the retrieved set and is easy to quote, you get cited.
Search Generative Experience (SGE)
Search Generative Experience (SGE) was Google's experimental name, used through 2023 and 2024, for the AI-generated answers shown at the top of Search and tested via Search Labs. It graduated into what Google now calls AI Overviews and AI Mode. So when you read older articles about SGE, they're describing the early version of the AI answers you see in Google today.
SearchGPT
SearchGPT was OpenAI's name for its AI search prototype, the one that answered questions with synthesized text plus a sources panel instead of a list of links. It has since been folded into ChatGPT as the feature now called ChatGPT Search. So if you read an older article about SearchGPT, it is describing what is now just ChatGPT Search. Same idea, new name, no separate product to go find.
Semantic search
Semantic search matches the meaning and intent behind a query instead of exact keywords. It turns text into vector embeddings (lists of numbers that capture meaning) and finds the closest matches in that space. It is how AI engines understand "comfy shoes for standing all day" even when those exact words never appear on the page, and it powers the retrieval step in AI search.
Share of voice (AI)
Share of voice (AI) is your brand's portion of all the mentions, citations, or recommendations across AI-generated answers for a set of questions, measured against your competitors. The basic math is your mentions divided by total mentions for everyone, times 100. So if AI engines name brands 200 times across your category prompts and you show up 50 times, your AI share of voice is 25%.
Vector search (embeddings)
Vector search finds results by turning text into numerical embeddings and retrieving the items whose vectors sit closest in meaning to your query. It is the engine room of semantic search and of the retrieval step that powers AI answers. Instead of matching the literal words you typed, it matches the concept behind them.
Zero-click search
A zero-click search is a search that ends without the user clicking any result, because the answer is already sitting on the results page (an AI Overview, a featured snippet, a knowledge panel). It has quietly become the norm for informational queries. That is why being the cited source inside the answer now matters more than ranking for a click that may never happen.