Skip to Main Content

AI Search Glossary

Plain-English definitions of the AI search vocabulary, from GEO and AEO to AI Mode, query fan-out and the crawler tokens that decide whether AI tools can read your site at all.

Paid Search GlossarySEO GlossaryWeb Development Glossary

What is AI search?

AI search is any search experience that generates a written answer instead of, or above, a list of links. In practice that means Google AI Overviews and AI Mode, ChatGPT search, Perplexity, and the answer features built into other assistants. The mechanics are the same across all of them: the tool breaks a question into parts, retrieves passages from pages it can reach, then writes a single response and cites some of the sources it used. Being cited is the new version of ranking.

Most of the confusion in this field is naming rather than substance. Generative search, answer engines and AI search all describe the same thing. AEO, answer engine optimisation, and GEO, generative engine optimisation, are two names for the work of getting cited in those answers, and you will also see AIO, LLMO, LLM SEO and AI SEO used for it. Google's own version was called the Search Generative Experience in 2023 and became AI Overviews in May 2024, with AI Mode added later as a separate conversational setting. The labels keep moving. The underlying job does not.

What changes against ordinary SEO is narrower than the acronyms suggest. Google's documentation on its AI features is explicit that there are no additional requirements to appear in AI Overviews or AI Mode, and no new file, AI text file or special schema.org markup to add. What does change is format and measurement. Search Console's generative AI performance report, live on all sites since 31 August 2026, gives impressions for AI Overviews and AI Mode and nothing else: no clicks, no click-through rate, no position, no queries. Clicks from those features land in the ordinary Web search type where they cannot be separated out. So the honest position is that you can see whether you are being surfaced, and you cannot yet see what it earned you.

The terms below define the vocabulary, from the acronyms through to the named crawler tokens that decide whether an AI tool can read your site at all. For the longer explanation of the shift rather than the definitions, start with what AI search means for your business and our guide to generative engine optimisation. How to appear in Google AI Overviews covers the practical side, and you can see how we approach it in our AI search optimisation services.

AEO (Answer Engine Optimisation)

The practice of structuring content so it appears as a direct answer in AI-powered search tools like Google AI Overviews, Perplexity, and ChatGPT. AEO focuses on clear, concise, answer-first writing that AI systems can extract and surface easily. In practice it describes the same work as GEO; which acronym you see is a matter of who is writing.

Agentic AI

AI systems that can autonomously plan and execute multi-step tasks - browsing the web, using tools, and taking actions on a user's behalf. As AI agents become more prevalent, they will increasingly research, compare, and recommend products and services without human search queries. Brands optimised for AI discovery today are better positioned for this shift.

AI Crawler

A bot that fetches web pages on behalf of an AI company, either to train a model, to build a search index, or to answer a question a user has just asked. The tokens are separate and each can be allowed or blocked individually in robots.txt. OpenAI runs GPTBot for model training, OAI-SearchBot to surface sites in ChatGPT search, OAI-AdsBot to check pages submitted as ads, and ChatGPT-User for user-initiated fetches. Anthropic runs ClaudeBot for training, Claude-SearchBot for search quality, and Claude-User for user-initiated fetches. Perplexity runs PerplexityBot for its search index, which it states is not used to crawl content for foundation models, and Perplexity-User for live user requests, which it says generally ignores robots.txt because a person triggered it. Blocking the training bot does not block the search bot, and that is the distinction most robots.txt files get wrong.

AI Mode

Google's conversational setting inside Search, which answers a question in full rather than returning a list of links. It is available in Australia in English. AI Mode works by query fan-out: it breaks the question into subtopics, issues a number of searches on your behalf, and synthesises the results into one response with links. Google has said early testers write queries nearly three times the length of a traditional search, so the phrasing that reaches AI Mode looks like a full sentence or a described situation rather than a two-word keyword.

AI Overviews

Google's AI-generated summaries that appear at the top of search results for many queries. AI Overviews pull from multiple sources to synthesise a direct answer. Google has brought a custom version of its Gemini 2.5 model into Search for both AI Overviews and AI Mode. Appearing in an AI Overview can drive significant brand visibility, even without a first-page ranking.

AI Search

The umbrella term for search experiences that generate a written answer instead of, or above, a list of links. It covers Google AI Overviews and AI Mode, ChatGPT search, Perplexity, and the answer features built into other assistants. The defining change is that the result is written rather than ranked, so the goal shifts from holding a position to being one of the sources the answer was built from. Generative search and answer engines are other names for the same thing.

Answer-First Format (BLUF)

A writing structure where the direct answer to a question is placed at the very beginning of a section, before any supporting explanation. BLUF stands for Bottom Line Up Front. This format is favoured by AI systems that extract answers from content, and by users who want quick answers.

Brand Entity

How AI systems 'understand' your business - not just your website, but the full picture assembled from your domain, business listings, social profiles, third-party mentions, and structured data. A strong brand entity means AI systems have consistent, accurate information about who you are, what you do, and where you operate. Entity confusion leads to hallucinations and missed citations.

ChatGPT

OpenAI's AI assistant, which answers questions from its training data and, through its search features, from live web results. OpenAI crawls for those two purposes with separate bots: GPTBot for model training and OAI-SearchBot for the search index behind ChatGPT's answers. As more people search inside ChatGPT rather than Google, being readable to the search bot is a meaningful part of modern SEO strategy.

Chunking

Splitting a page into smaller passages so a retrieval system can pull the single part that answers a question rather than the whole document. Retrieval-augmented systems work on these chunks, not on whole pages, which is why a page that buries its answer in the middle of a long undifferentiated section performs worse than one where each heading owns a single question and answers it in the first line underneath.

Citation

When an AI system attributes its answer to a specific source - linking to or naming the website it drew from. Being cited by ChatGPT, Perplexity, or Google AI Overviews is the AI search equivalent of a first-page ranking. Citations are driven by content quality, structured formatting, brand authority, and presence on platforms AI engines trust.

Conversational Search

Search queries written as natural language questions or multi-turn conversations, rather than keywords. As AI search matures, users increasingly ask full questions like 'what is the best Google Ads agency in Perth for a small business'. Content structured around natural questions and direct answers performs better in conversational search contexts.

Crawlability (for AI)

Whether AI systems can access and process your website's content. The AI crawlers that respect robots.txt are blocked by name, so a blanket rule written for Googlebot will not necessarily cover them, and a rule aimed at stopping model training can accidentally remove you from an AI search index as well. Auditing which named bots your robots.txt actually allows is now part of a technical SEO review.

E-E-A-T

Experience, Expertise, Authoritativeness, and Trustworthiness - the framework Google uses to evaluate content quality. AI systems tend to favour content that demonstrates genuine expertise and real-world experience, making E-E-A-T signals increasingly important for AI search visibility.

Entity SEO

Optimising how your business, people, and topics are represented as entities in search and AI systems - beyond just keywords. Entity SEO involves building consistent structured data, maintaining accurate business listings, earning third-party mentions, and ensuring your brand is understood as a distinct, trustworthy entity. Strong entity signals improve both traditional rankings and AI citation rates.

Gemini

Google's family of large language models, available as the standalone Gemini app and used across Google's products. Google has stated that it brought a custom version of Gemini 2.5 into Search to power both AI Mode and AI Overviews. It matters for visibility that one company controls both the ranking system and the model writing the answer, because the content signals that earn an ordinary ranking are also the ones feeding the generated response.

Generative AI Performance Report

The Search Console report showing how often links to your site were shown inside Google's generative AI features. Google rolled it out to all sites worldwide on 31 August 2026, and it covers AI Overviews and AI Mode. Read its limits carefully: it reports impressions only, with no clicks, no click-through rate, no average position and no query data. Impressions can be broken down by page, country, device, date, and whether the search was text-based or multimodal. Clicks from AI features are still counted, but inside the ordinary Web search type of the main performance report, where they cannot be separated from classic organic clicks.

Generative Search

Another name for AI search, emphasising that the answer is generated by a language model rather than retrieved from an index. Generative search, AI search and answer engines are used interchangeably in practice. Google's own version began as the Search Generative Experience in 2023 and was renamed AI Overviews in May 2024, which is why older writing uses a label the product no longer carries.

GEO (Generative Engine Optimisation)

The practice of optimising content to appear in AI-generated search results. GEO extends traditional SEO by focusing on how AI systems select, synthesise, and attribute information, including structured formatting, factual density, and clear sourcing. GEO and AEO describe the same work under different names.

Google-Extended

The robots.txt token controlling whether Google may use your content to train future Gemini models and to ground answers in Gemini Apps and Vertex AI. Google documents that Google-Extended does not impact a site's inclusion in Google Search, and is not used as a ranking signal in Google Search. The consequence catches people out: disallowing it is not an opt-out from AI Overviews or AI Mode, because those are part of Search and are governed by the ordinary indexing and snippet controls instead.

Grounding

The process of connecting an AI's response to real-world, verifiable information - typically by retrieving live web content before generating an answer. Grounded AI responses are more accurate and up-to-date than ungrounded ones. Google AI Overviews and Perplexity both use grounding, which is why well-structured, factually accurate web content can be cited even by AI systems with a training cutoff in the past.

Hallucination

When an AI model generates information that is factually incorrect or fabricated, presented with apparent confidence. Hallucinations are a known limitation of LLMs. Well-sourced, clearly structured content on your website can help AI systems cite accurate information about your business.

Knowledge Graph

Google's database of entities - people, places, organisations, and concepts - and the relationships between them. When Google's Knowledge Graph has a strong entity record for your business, it can answer questions about you directly in search results and improve your chances of appearing in AI Overviews. Business listings, schema markup, and consistent NAP data all contribute to your Knowledge Graph presence.

LLM (Large Language Model)

An AI model trained on vast amounts of text data to understand and generate human-like language. LLMs power tools like ChatGPT, Gemini, and Claude. They are increasingly being used to answer search queries directly, shifting how users discover information online.

llms.txt

A proposed file, placed at the root of a domain, listing a site's most important pages in a format convenient for language models. It has publishers but no confirmed consumer. Asked in January 2026 whether llms.txt files on Google's own properties amounted to an endorsement, Google's John Mueller answered, to be direct, no. Google's documentation on AI features separately states you do not need to create new machine readable files or AI text files to appear in them. Some AI vendors do reference the format in their own developer documentation. It is cheap to publish and speculative to rely on, so treat it as the latter rather than as a ranking factor.

Multimodal AI Search

AI search that processes and understands multiple types of input - text, images, voice, and video - rather than text queries alone. Google Lens and ChatGPT's image analysis are early examples. Search Console now reports multimodal web search separately from text-based web search, which is the first time the split has been measurable. Multimodal search is expanding what 'discoverability' means: image alt text, video transcripts, and structured product data are all increasingly relevant.

Perplexity

An AI-powered search engine that provides direct, cited answers to queries by searching the web in real time. It crawls with PerplexityBot for its own search index, which Perplexity states is not used to crawl content for AI foundation models, and fetches pages live as Perplexity-User when a user asks a question. Perplexity is growing in popularity as an alternative to Google, particularly among technical and research-oriented users.

Preview Controls

The directives limiting how much of a page Google may show in a result, and the documented way to limit what appears in AI features: nosnippet to suppress the snippet, data-nosnippet to suppress a specific element, max-snippet to cap the character count, and noindex to stay out of results altogether. Google points at these rather than at any AI-specific tag. The trade-off runs both ways, because there is no separate switch for AI Overviews, so restricting what an AI feature can quote restricts your ordinary results by the same amount.

Query Fan-Out

The process AI search engines use to break a complex user query into multiple sub-questions, search for answers to each, and synthesise a comprehensive response. Google describes AI Mode as working this way. Content that anticipates and answers related sub-questions is more likely to be surfaced across multiple query variations.

RAG (Retrieval-Augmented Generation)

A technique where an AI retrieves relevant information from external sources before generating a response. Google AI Overviews and Perplexity both use RAG - they search the web, retrieve relevant content, then generate a synthesised answer. Retrieval happens at passage level rather than whole-page level, so well-structured content is more likely to be retrieved and used.

SGE (Search Generative Experience)

The name Google gave its first generative answer experiment, introduced at Google I/O in 2023 and renamed AI Overviews when the feature launched properly at Google I/O in May 2024. The term is now legacy. If a strategy document still talks about optimising for SGE, it predates the product it is describing.

Structured Data

Code added to a page that explicitly describes its content in a format machines can understand. Structured data (typically Schema.org JSON-LD) helps both search engines and AI systems accurately understand who you are, what you do, and what your content covers. Worth being precise about what it does: Google states there is no special schema.org type required to appear in AI Overviews or AI Mode, so the value is in describing your business and content accurately, not in a markup shortcut to AI visibility.

Training Cutoff

The date after which an AI model has no knowledge from its training data. Events, businesses, or content created after the training cutoff are invisible to the model unless it uses RAG (real-time web retrieval). This is why keeping your website content current and web-accessible matters - AI tools that browse the web can surface recent content regardless of their training cutoff.

Zero-Click Search

A search result where the user gets their answer directly on the results page - via AI Overviews, featured snippets, or knowledge panels - without clicking through to any website. Zero-click searches are increasing as AI search matures, making brand visibility in these results an important goal even without a click.

AI search, explained.

AI search is any search experience that writes an answer rather than returning only a list of links. The main ones are Google AI Overviews and AI Mode, ChatGPT search, Perplexity, and the answer features inside Gemini and other assistants. Mechanically they all do the same three things: break the question into parts, retrieve passages from pages they can reach, then generate a single response and cite some of the sources behind it. The practical consequence for a business is that visibility stops being a position on a page and becomes whether you are one of the sources the answer was built from. You can rank first in the blue links and still be absent from the answer above them, and you can be cited in the answer without ranking on page one at all.
SEO is optimising to rank in a list of results. AEO, answer engine optimisation, and GEO, generative engine optimisation, are two names for optimising to be cited inside a generated answer. In practice they describe the same work and the choice between them is branding rather than method. You will also see AIO, LLMO, LLM SEO and AI SEO used for it. The more useful question is how much of the work is genuinely new, and the honest answer is less than the vocabulary implies: Google documents that there are no additional requirements to appear in AI Overviews or AI Mode and no special schema.org structured data to add. What does change is format and measurement. Answer the question in the first sentence under each heading so a passage can be lifted cleanly, keep your entity consistent across the web so you are attributed correctly, and accept that the reporting on the result is much thinner than it is for organic rankings.
Yes. Generative search, AI search and answer engines are interchangeable labels for search that generates a written response instead of only ranking documents. The naming has shifted more than the product has. Google introduced its version as the Search Generative Experience, or SGE, at Google I/O in 2023, renamed it AI Overviews when the feature launched properly at Google I/O in May 2024, and later added AI Mode as a separate conversational setting inside Search. That history is why material written in 2023 and 2024 talks about optimising for SGE, a product name Google no longer uses. If a strategy document still says SGE, it predates the thing it is describing.
No. Google is explicit in its documentation on AI features in Search: you do not need to create new machine readable files, AI text files or markup to appear in them, and there is no special schema.org structured data you need to add. On llms.txt specifically, which is a proposed file listing a site's key pages for language models, Google's John Mueller was asked in January 2026 whether llms.txt files on Google's own properties amounted to an endorsement and answered, to be direct, no. Nothing stops you publishing one, it costs very little, and some AI vendors reference the format in their own developer documentation. Just do not treat it as a ranking factor, and do not let it displace the things that do matter, which are ordinary crawlability, structured data that describes your business accurately, and content that answers the question in its first line.
Search Console has a generative AI performance report, which Google rolled out to all sites worldwide on 31 August 2026. It shows how many times links to your site were shown inside a generative AI feature, and it covers AI Overviews and AI Mode. The limits are severe and worth knowing before you build reporting on it. The report gives impressions only: no clicks, no click-through rate, no average position and no query data. You can break those impressions down by page, country, device, date, and whether the search was text-based or multimodal, and that is the whole report. Clicks from AI features are still counted, but they are folded into the ordinary Web search type in the main performance report, where they cannot be separated from classic organic clicks. So you can establish that you are being surfaced, and you cannot yet establish what it earned you.
Yes, in English. Google announced the Australian rollout on its own Australian blog, and AI Mode is a setting inside Search rather than a separate product. It works by query fan-out: it breaks your question into subtopics, issues a number of searches behind the scenes, and synthesises the results into one conversational answer with links through to sources. The behavioural shift is the part worth planning around. Google has said early testers write queries nearly three times the length of a traditional search, which means the phrasing people reach AI Mode with looks like a full sentence or a described situation rather than a two-word keyword. Content built only around short head terms tends not to match those questions. Content organised as explicit questions with direct answers underneath does.
Largely yes, because the crawlers are separated by purpose and each can be allowed or blocked by name in robots.txt. OpenAI runs GPTBot for model training, OAI-SearchBot to surface sites in ChatGPT search, and ChatGPT-User for pages fetched because a user asked. Anthropic runs ClaudeBot for training, Claude-SearchBot for search quality, and Claude-User for user-initiated fetches. Perplexity runs PerplexityBot for its search index, which it states is not used to crawl content for foundation models, and Perplexity-User for live user requests, which it says generally ignores robots.txt because a person triggered it. Google splits it the same way: Google-Extended controls whether your content may be used to train future Gemini models and to ground answers in Gemini Apps and Vertex AI, and Google documents that it does not impact a site's inclusion in Google Search and is not used as a ranking signal. The catch is that AI Overviews and AI Mode are part of Search, so disallowing Google-Extended does not opt you out of them. The only levers there are the ordinary indexing and snippet controls, which cut your normal results by the same amount.
Some, though not in a way the current tooling lets you measure on your own site. Search Console reports generative AI impressions with no clicks attached, and the clicks that do come from AI features are mixed into the ordinary Web search metrics, so a site owner cannot isolate them. What can be said with confidence is the direction. Answers that resolve a question in full produce zero-click searches, where the user has what they came for and never visits a site, and the queries most exposed to that are definitional and informational ones. Commercial and comparison queries, where the searcher still has to choose a supplier and wants to check them, hold up better. The strategic response is not to abandon informational content, because that is often what gets you cited and keeps you in the consideration set. It is to stop valuing that content purely on sessions, and to make sure the pages that have to convert are the ones answering commercial questions.

Want to show up in AI search results?

Book a free 20-minute strategy call and we will review your current visibility and map out what it takes to appear in AI Overviews and LLM responses. Or explore our AI search optimisation services.

Get in touch