AI search has its own vocabulary, and much of it is new, overlapping, or coined by vendors who sell tools. This GEO glossary sorts the terms out in plain English. It covers the optimization labels (GEO, AEO, LLMO, AIO, and "SEO for AI"), the engine mechanics, the metrics, and the crawlers and files, and it links each term to the deeper guide that owns it.
Neutrality is the point here. A glossary that hypes one vendor's word over another is less useful and less trustworthy, so every entry below sticks to what a term means and where it came from. Where a term is a marketing label rather than a research term, this glossary says so plainly.
The optimization terms (GEO, AEO, LLMO, AIO)
The four terms below all describe roughly the same work: getting AI answers to cite or name your business. They differ mainly in origin and shade of meaning, not in practice.
Generative Engine Optimization (GEO)
Generative Engine Optimization (GEO) is the practice of shaping content so AI answer engines cite or name a business in their generated replies. GEO is an academic term, coined by Aggarwal and coauthors in a Princeton-led paper (arXiv 2311.09735, published at KDD 2024), not a vendor invention. Note the caveat: that paper's roughly 30 to 40 percent visibility lift is a maximum for a few specific methods measured on a simulated engine, not a general guarantee for live ChatGPT or Perplexity.
Answer Engine Optimization (AEO)
Answer Engine Optimization (AEO) is optimizing content so answer engines like ChatGPT, Perplexity, and Google AI Overviews pull it out as a direct answer. AEO is an industry and vendor term with no single authoritative origin, and in plain use AEO overlaps heavily with GEO.
Large Language Model Optimization (LLMO)
Large Language Model Optimization (LLMO) is a vendor term for shaping content and off-site signals so large language models cite or recall a brand. LLMO has no single author. One ambiguity is worth flagging: some technical sources use LLMO to mean optimizing the models themselves, such as pruning or fine-tuning, not marketing visibility.
AI Overview Optimization (AIO)
AI Overview Optimization (AIO) is a vendor term for optimizing specifically for Google AI Overviews. AIO is an ambiguous acronym, because the same letters are also used for Artificial Intelligence Optimization, which is a broader idea. Check which meaning a source intends before you act on an AIO claim.
How the AI engines work
These terms describe the machinery behind AI answers: what a generative engine is, how it splits a question, and how it pulls in outside facts.
Generative engine
A generative engine is a search or answer system that uses a large language model to pull from many sources and write a single synthesized answer. The term generative engine was formalized in the same Princeton-led GEO paper (Aggarwal and coauthors, arXiv 2311.09735).
Answer engine
An answer engine is a system that returns one direct, written answer to a question instead of a ranked list of blue links. Answer engine is an industry term that overlaps with generative engine.
Query fan-out
Query fan-out is Google's technique of breaking one question into several smaller searches, running them at the same time, and then combining the results into a single answer. Google documents query fan-out for both AI Mode and AI Overviews (Google I/O 2025).
Retrieval-Augmented Generation (RAG)
Retrieval-Augmented Generation (RAG) is an architecture that pairs a language model with a live retrieval step, so the model pulls in outside documents before it answers. RAG was introduced by Lewis and coauthors at Facebook AI Research, University College London, and New York University (arXiv 2005.11401, NeurIPS 2020).
AI Overviews
AI Overviews are Google's AI-written summaries that appear above the normal search results. Google launched AI Overviews at Google I/O in May 2024 for United States users. Our guide covers how to rank in Google AI Overviews.
AI Mode
AI Mode is Google's separate, conversational search experience that uses query fan-out and the Gemini model. Google opened AI Mode in Search Labs in March 2025 and rolled it out to all United States users in May 2025. See how to show up in Google AI Mode.
Metrics and signals
These terms describe what you try to earn from AI answers and how people try to measure it.
AI citation
An AI citation is an attributed reference, usually a link or a named source, that an AI answer includes to credit where its information came from. Earning citations is the goal of GEO. Our guide covers how to get cited by AI.
Brand mention (unlinked)
A brand mention (unlinked) is a plain-text reference to a business with no hyperlink attached. Unlinked mentions matter because vendor research (Ahrefs) finds branded mentions correlate with AI visibility, and in this setting a mention without a link appears to behave much like one with a link. You can track brand mentions in AI to see where yours show up.
AI share of voice
AI share of voice is a metric that estimates how often, and how prominently, a brand shows up in AI answers compared with its competitors. AI share of voice is a vendor metric with no standard definition, and different tools such as Semrush and Profound measure it in different ways. Our guide covers how to measure AI visibility without over-trusting one number.
Crawlers and files
These terms describe the files and bots that decide whether AI can read your site at all. If you get one thing right first, make sure the crawlers are not blocked. See AI crawler access.
llms.txt
llms.txt is a proposed file, placed at a site's root, meant to give AI models a clean, curated index of a site's key content. Jeremy Howard of Answer.AI proposed llms.txt on September 3, 2024. llms.txt is only a proposal, not a ratified standard, and its adoption and value are disputed. We cover the evidence on whether llms.txt works.
GPTBot
GPTBot is OpenAI's web crawler, named by the user-agent token GPTBot, used to gather public web pages to train OpenAI's models. OpenAI introduced GPTBot in August 2023, and you control it through robots.txt. GPTBot is separate from OAI-SearchBot and ChatGPT-User, so blocking GPTBot affects training, not necessarily whether ChatGPT search cites you.
The GEO acronym family, side by side
GEO, AEO, LLMO, and AIO are best treated as one family of near-synonyms rather than four separate disciplines. The table below lines up what each one optimizes for, who coined it, and what else it is called. For how this whole family differs from classic search, see GEO versus SEO.
| Term | What it optimizes for | Origin / who coined it | Also called |
|---|---|---|---|
| GEO | Being cited or named in AI-generated answers | Academic. Aggarwal and coauthors, Princeton-led (arXiv 2311.09735, KDD 2024) | AEO, LLMO, "SEO for AI" |
| AEO | Being pulled out as the direct answer by answer engines | Industry / vendor. No single author | GEO, LLMO |
| LLMO | Being cited or recalled by large language models | Vendor. No single author | GEO, AEO |
| AIO | Showing up in Google AI Overviews | Vendor. No single author | AI Overview Optimization |
One honest note on the table. AIO is an ambiguous acronym, since the same letters also stand for Artificial Intelligence Optimization, so read the source before you assume which one is meant. And the "also called" column is not a promise that the terms are identical in every writer's usage, only that they are treated as synonyms in most marketing.
How to use this glossary
You get the most from this GEO glossary by reading each entry as a neutral definition first and a link to a deeper guide second. Here is the simple order.
- Find your term. Use the section headings to jump to the optimization labels, the engine mechanics, the metrics and signals, or the crawlers and files.
- Read the neutral definition. Each entry says what the term means in one or two plain sentences, with no hype and no sales angle.
- Follow the link to the deep guide. Most terms link to a full page on the site that goes further than a one-line definition.
- Check the origin before you trust a vendor claim. If a term is flagged as vendor-coined, treat bold percentage claims about it with care and look for a primary source.
The last step is the one that saves you money. A lot of AI-search advice leans on numbers that come from the same vendors selling the fix. Knowing which terms are academic and which are marketing labels helps you weigh those claims.
Origin is not trivia. GEO is a research term with a paper behind it, while AEO, LLMO, and AIO are marketing labels that vendors shaped. That does not make the vendor terms wrong, but it does mean a bold statistic tied to one of them deserves a look at the source. When a term has no author, a claim about it has no author either.
Common questions about GEO and AI-search terms
Is GEO the same as AEO?
GEO and AEO are close synonyms in everyday marketing use, and most people treat them as the same work. The difference is origin. GEO (Generative Engine Optimization) is an academic term from a Princeton-led paper, while AEO (Answer Engine Optimization) was popularized by SEO vendors with no single author. For how this family compares with classic SEO, see our guide on GEO versus SEO.
Who invented the term GEO?
The term GEO was coined by Pranjal Aggarwal and coauthors in a Princeton-led research paper titled "GEO: Generative Engine Optimization" (arXiv 2311.09735), published at the KDD 2024 conference. GEO is therefore an academic term, not a vendor coinage. One caution: the paper's roughly 30 to 40 percent visibility lift is a maximum for a few specific methods on a simulated engine, not a general promise.
Is llms.txt an official standard?
No. llms.txt is a proposed file, suggested by Jeremy Howard of Answer.AI in September 2024, and it has not been ratified by any standards body. Adoption and real-world value are still disputed, and no major AI engine is on record using it. Our full guide covers whether llms.txt works.
What is the difference between GPTBot and OAI-SearchBot?
GPTBot and OAI-SearchBot are two separate OpenAI crawlers with different jobs. GPTBot gathers public pages to train OpenAI's models, while OAI-SearchBot handles pages for ChatGPT's search feature. Blocking GPTBot in robots.txt limits training use, but it does not by itself stop ChatGPT search from citing you. Our AI crawler access guide shows how to check which bots you allow.
Is AI share of voice a real metric?
AI share of voice is a real thing to measure, but it is not a standard metric. AI share of voice estimates how often a brand appears in AI answers next to competitors, and different vendors such as Semrush and Profound define and measure it differently. Treat any single number as one tool's estimate, not a fixed score. Our guide covers how to measure AI visibility honestly.