Knowledge base
AI Search Glossary: 30 Terms From GEO to GPTBot, in Plain English
One-line definitions, three reading paths and a map of how an AI answer gets built. Sourced to the companies that own each product, and dated.
AI search glossary: what this knowledge base covers
This AI search glossary is Digital Magma’s knowledge base of 30 terms that explain how AI tools find, read and cite websites, from GEO and AEO to query fan-out, RAG and GPTBot. Each term gets a plain-English definition here, and each full entry adds how it works, why it matters and what to do.
Vendors now use GEO, AEO and LLMO as if they were three products.
Robots.txt advice often mixes up training bots with search bots. The glossary exists to say what each term means and who defined it.
Key takeaways
- The 30 terms fall into five groups: strategies, engines, mechanics, crawlers and SEO foundations.
- Three reading paths, for owners, marketers and developers, give a sensible order.
- The answer map follows an AI reply from crawl to citation, one term per step.
- Facts come from the owner of each product or standard, and every entry is dated.
- Google’s July 2026 guide treats AEO and GEO work as SEO, and this glossary follows that view.
Where to start: reading paths for owners, marketers and developers
Pick the AI search glossary path that fits your job, and read the terms in that order.
Business owners
Start with AI SEO, then AI Overviews and AI citation. Finish with AI share of voice, so you know what an agency report should show.
Read GPTBot and Google-Extended before anyone edits your robots.txt.
Marketers and content teams
Begin with GEO and AEO, then query fan-out. Move on to schema markup, E-E-A-T and entity SEO.
These shape how a page is written and structured.
Developers
Start with AI crawler and robots.txt for AI. Then read GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot.
End with llms.txt, RAG and grounding to see what happens after the fetch.
Short on time? Digital Magma suggests two entries: AI SEO and AI Overviews. Together they cover most of what a first agency call will throw at you.
How an AI answer is built, term by term
An AI search answer is built in a fixed order, and each step has a glossary term attached.
- Crawl. An AI crawler fetches the page, if robots.txt allows it.
- Index and understand. The engine stores the page. Schema markup and entity SEO help it know who and what the page is about.
- Fan out. The question is split into related searches through query fan-out.
- Retrieve. Retrieval-augmented generation pulls the best matching passages.
- Ground and write. A large language model writes the reply, with grounding tying it to the sources.
- Cite. Some sources earn an AI citation, a visible link.
- Measure. AI share of voice tracks how often you are named across many prompts.
Google describes the middle steps in its July 2026 guide. It names RAG, which it says is also called grounding, and query fan-out as the techniques behind AI Overviews and AI Mode.
Digital Magma audits sites in this same order, from crawl access to citations.
AI SEO strategies and metrics: GEO, AEO, LLMO and AI share of voice
AI SEO strategies set the goal, and these six terms name the disciplines and the two ways success is counted.
- AI SEO
- The umbrella for being found, described correctly and cited by AI search tools. It builds on technical and content SEO.
- Generative engine optimization (GEO)
- Getting content cited by engines that retrieve sources and then write. Named in a 2023 Princeton and IIT Delhi paper.
- Answer engine optimization (AEO)
- Making one passage the lifted answer, in featured snippets, voice replies and AI summaries.
- LLM optimization (LLMO)
- Influencing what language models say about a brand when they answer from memory, mostly through consistent facts across the web.
- AI citation
- A linked source shown with an AI answer. A mention without a link is a different, weaker signal.
- AI share of voice
- Your share of brand mentions across a fixed prompt set, against named rivals. Tools and agencies calculate it, and no AI platform publishes it.
AI search engines and products: AI Overviews, AI Mode, ChatGPT search, Perplexity, Copilot and Gemini
AI search now runs inside six main products, each with its own crawler, index and rules.
- Google AI Overviews
- AI summaries above some Google results, with supporting links. US launch in May 2024.
- Google AI Mode
- A conversational Search mode for follow-ups and comparisons, which Google says may use query fan-out.
- ChatGPT search
- OpenAI’s web search inside ChatGPT, launched on 31 October 2024. It depends on OAI-SearchBot access.
- Perplexity
- An answer engine that searches the web for each question and links what it used. Its search crawler is PerplexityBot.
- Microsoft Copilot Search
- Microsoft’s AI answers, tied to Bing. Bing Webmaster Tools reports its citations in AI Performance.
- Gemini
- Google’s model family and app. The Google-Extended token governs training and grounding use in Gemini Apps.
AI search mechanics: query fan-out, RAG, grounding and citations
AI search mechanics explain why one page gets quoted and another is ignored.
- Query fan-out
- A burst of related searches the model runs at once to cover a question’s subtopics.
- Retrieval-augmented generation (RAG)
- Fetching relevant documents first, then handing them to a model to write from, so answers can use current sources.
- Grounding
- Tying a model’s reply to material supplied at answer time. Google uses the word as another name for RAG.
- Large language model (LLM)
- A model trained on huge amounts of text to read and write language. GPT, Claude and Gemini are examples.
- Semantic search
- Matching meaning rather than exact words, which is why a page can be cited for phrasing it never uses.
AI crawlers and access controls: GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended
AI crawlers decide what each company can read from your site, and robots.txt is how you answer them.
- AI crawler
- A bot from an AI company that fetches pages for training, search or a user’s live request, under its own user-agent token.
- GPTBot
- OpenAI’s training crawler. Blocking it leaves ChatGPT search untouched, since OpenAI says each bot setting is independent.
- OAI-SearchBot
- OpenAI’s search crawler. Opted-out sites are not shown in ChatGPT search answers.
- ClaudeBot
- Anthropic’s training crawler. Anthropic runs Claude-SearchBot and Claude-User separately for search and user requests.
- PerplexityBot
- Perplexity’s search crawler, which Perplexity says is not used to collect content for foundation models.
- Google-Extended
- A robots.txt token for Gemini training and Gemini Apps grounding. Google says it has no effect on Search inclusion or ranking.
- llms.txt
- A proposed Markdown guide to a site for AI agents. Optional, and ignored by Google Search.
- Robots.txt for AI
- Robots.txt groups that allow or block AI crawlers by token, under the RFC 9309 standard. Compliance is voluntary.
SEO foundations AI systems rely on: schema, entities, Knowledge Graph and E-E-A-T
SEO foundations still carry AI search, and these five shape how machines understand who you are.
- Schema markup
- Structured data, usually JSON-LD, that labels people, products and organisations on a page. It must match the visible text.
- Entity SEO
- Making sure engines recognise your business, people and products as distinct things with consistent facts.
- Knowledge Graph
- Google’s database of entities and the links between them, which feeds panels and answers.
- E-E-A-T
- Experience, expertise, authoritativeness and trust, from Google’s rater guidelines. Google’s helpful content guide says it is not a ranking factor in itself.
- Speakable schema
- Markup for text suited to reading aloud. Google’s support is a beta for English news on Google Assistant.
How this AI SEO glossary is sourced and kept current
This AI SEO glossary is maintained by Ankit Sachan for Digital Magma. It takes each fact from the company or group that owns the product or standard, and names that publisher in the text.
Primary sources used
- Google Search Central, Optimizing your website for generative AI features on Google Search, updated 10 July 2026.
- Google Search Console help, Generative AI performance report, rolled out to all sites on 31 August 2026.
- OpenAI, Overview of OpenAI Crawlers: GPTBot, OAI-SearchBot, ChatGPT-User and OAI-AdsBot, with about 24 hours for search changes.
- Anthropic Help Center, crawler article, updated 7 April 2026.
- Perplexity, Perplexity Crawlers, up to 24 hours for changes.
- Jeremy Howard, The /llms.txt file, v2, first published 3 September 2024 and modified 10 August 2026.
- Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024. Its percentages appear only as the authors’ benchmark results.
A statistic appears only when a named source published it. Inferences are labelled as Digital Magma’s view.
Change log
- 21 October 2026: added reading paths, the answer map, Google’s July 2026 guide and the Search Console report. All 30 terms now carry a definition.
- 15 October 2026: first draft with eight full entries.
Each term becomes a link as its full entry goes live. Spot an error? Send it with the source through our contact page.
AI search glossary: common questions
AI search glossary questions readers send us, with short answers.
What is the difference between AI SEO, GEO and AEO?
AI SEO is the umbrella for visibility in AI search. GEO aims to be one of several sources an AI answer cites.
AEO aims to be the single passage lifted as the answer. Google’s July 2026 guide calls all of this SEO.
Which AI search terms should a business owner learn first?
Start with AI SEO, AI Overviews and AI citation. They explain what AI search is, where Google shows it and how to tell if you appear. Add GPTBot and Google-Extended before changing robots.txt.
What are the most important AI search terms in 2026?
Query fan-out, RAG and grounding explain how answers are built. OAI-SearchBot and Google-Extended decide access.
The Generative AI performance report, new for all sites in August 2026, is now the main Google measurement term to know.
Do Google AI Overviews need special optimization?
Google says they need none. A page must be indexed, snippet-eligible and included in Search generative AI features in Search Console. Beyond that, normal SEO applies. The planned AI Overviews entry covers controls and tracking.
Is llms.txt required to appear in AI answers?
It is not required. Google Search ignores llms.txt, so it neither helps nor harms there.
The v2 proposal says thousands of sites publish one and some agents read it, so it can help with other tools.
Which AI crawlers should a website allow?
Digital Magma’s default advice for businesses that want AI visibility is to allow the search crawlers: Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot and PerplexityBot. Training crawlers such as GPTBot, ClaudeBot and Google-Extended are a separate business decision.
How often is this glossary updated?
Entries are reviewed whenever the primary documentation changes, and each shows its last-updated date. The change log above records each revision of this index.
Where can I ask a question this glossary does not answer?
Send it through the contact page. Buyer questions will also be answered in a planned Ask the Expert section.
AI search help from Digital Magma
Digital Magma applies this AI search glossary in its client work, from crawler audits to citation tracking.
To see it applied to your site, read our AI SEO services page or ask for a free SEO audit. No ranking or AI citation is guaranteed.
Call +91 79823 19697, email contact@digitalmagma.com or message on WhatsApp (Mon–Sat, 9 am–9 pm IST). You can read our Google reviews first.