The language of AI search.
Definitions of common AI search terms.
A
- AEO
- Short for Answer Engine Optimization. Work on content and technical foundations to help systems that provide direct answers find, interpret, and use a business’s information.
- AI Assistant
- An application that uses AI to respond to requests, provide information, or help complete tasks. It may combine language models with tools such as web search, document retrieval, and other software.
- AI Crawler
- An automated program that visits websites and collects information for an AI service.
- AI information page
- An official website page that brings together key facts about a business, with links to supporting information. It gives readers and AI assistants a concise reference for the business’s services, locations, leadership and contact details.
- AI Mode
- A Google Search feature that uses AI to answer questions and lets people ask follow-up questions. Answers include links to relevant websites.
- AI Overviews
- AI-written summaries that appear in some Google search results, with links to websites where people can learn more.
- AI Recommendation
- When an AI assistant suggests a business, product, or service as relevant to a need or question. A mention or citation alone does not necessarily constitute a recommendation.
- AI Visibility
- The extent to which a business appears and is represented in AI-generated answers. Assessment can consider mentions, citations, recommendations, and accuracy across a defined set of questions and platforms.
- AIO
- An abbreviation used for both AI Optimization and Google’s AI Overviews. Its meaning depends on the context.
- Authority Signals
- Evidence that can help establish a source’s expertise, credibility, or reputation, such as relevant credentials, independent references, and reviews. The signals used and their importance vary across search systems.
C
- ChatGPT-User
- OpenAI’s user-agent identifier for certain visits to websites in response to a person’s request in ChatGPT.
- Claude-SearchBot
- Anthropic’s crawler that finds website content for Claude’s search results.
- Claude-User
- Anthropic’s user-agent identifier for certain visits to websites in response to a person’s request in Claude.
- ClaudeBot
- Anthropic’s crawler that collects website content that may be used to train its AI models.
D
- Data Layer
- The part of a software system that manages and provides access to data. In website analytics, the term also refers to a structure that passes page and user-action information to measurement tools.
E
- Entity Disambiguation
- Identifying which person, organization, place, or other entity a name or reference refers to in a particular context. This helps systems distinguish businesses and other entities with similar names or descriptions.
G
- Generative AI Platform
- A service or software environment that provides access to generative AI models and their capabilities. It may offer a user interface, developer tools, or both for producing text, images, audio, and other outputs.
- GEO
- Generative Engine Optimization: content and technical work to improve the visibility and representation of organizations, products, or sources in answers produced by generative search systems. Its scope overlaps with AEO and LLMO.
- Google-Extended
- A robots.txt control for whether Google may use crawled content to train Gemini models and support certain Gemini features that draw on web information. It is not a separate crawler and does not control Google Search inclusion or ranking.
- GPTBot
- OpenAI’s crawler that collects website content that may be used to train its AI models.
- Grounding
- Basing an AI response on specified source information, such as retrieved documents or database records. Retrieval-augmented generation is one method of supplying that information to the model.
H
- Hallucination
- Output from an AI model that presents false or unsupported information as factual.
J
- JSON-LD
- A way to express linked data, describing things and their relationships, in the JSON data format. Websites commonly use JSON-LD to publish schema.org structured data.
K
- Knowledge Graph
- A structured representation of entities, such as people, organizations, or products, and the relationships between them. It organizes information as a network that software can query and use to connect related facts.
L
- Large Language Model (LLM)
- A language model built from a large neural network and trained on extensive datasets to learn patterns in language. It uses those patterns and the context it receives to generate, summarize, translate, or classify text.
- LLMO
- Short for Large Language Model Optimization. In AI search marketing, the term describes content and technical work to improve how a business is represented in answers from language models.
- llms.txt
- A proposed, optional Markdown file that gives AI tools an overview of a website, guidance, and links to relevant content.
O
- OAI-SearchBot
- OpenAI’s crawler that finds website content for ChatGPT’s search results.
P
- Perplexity-User
- Perplexity’s user-agent identifier for certain visits to websites in response to a person’s request.
- PerplexityBot
- Perplexity’s crawler that finds website content for its search results.
- Platform Parity
- An informal term for a comparable level of visibility across selected AI platforms. Comparisons depend on the questions, measures, and time period selected.
Q
- Query Alignment
- An informal term for how closely content addresses the intent and information needs behind a search query.
R
- RAG (Retrieval-Augmented Generation)
- A method that retrieves relevant information from external sources and supplies it to a generative AI model as context for generating an answer. Sources can include a domain-specific database, a collection of documents, or web content.
- robots.txt
- A website file that gives automated crawlers instructions about which web addresses to request or avoid. The rules rely on crawlers following them and do not block access or protect private information.
S
- Schema Markup
- Structured data that uses schema.org’s shared vocabulary to describe things such as organizations, products, services, and their relationships. It can be written in formats such as JSON-LD, Microdata, or RDFa.
- Structured Data
- Information organized using a defined format or model, such as named fields, data types, and relationships. Schema markup is one application of structured data, alongside records in databases, feeds, and other systems.