Agentic Search
에이전틱 검색
Search in which AI autonomously expands queries and iterates through multiple stages of retrieval and evaluation to conduct detailed research.
Explore AI search, GEO and AEO terms with clear definitions, practical business applications and guidance on implementation and related concepts.
에이전틱 검색
Search in which AI autonomously expands queries and iterates through multiple stages of retrieval and evaluation to conduct detailed research.
에이전틱 웹
A web environment in which AI agents autonomously browse, read data and execute tasks alongside human users.
AI 에이전트
An intelligent AI system that autonomously browses the web, invokes tools, reasons and synthesizes results to achieve a given goal.
AI 인용
A source identified by a generative AI system through a footnote or link as factual support for a statement in its answer.
AI 크롤러
A dedicated bot that collects web content for generative AI training data or real-time search-based answers.
AI Mode
A search engine mode that replaces a results-list interface with conversational AI answers and interactive exploration.
AI Overviews
A Google Search feature that synthesizes multiple web sources into an AI-generated summary with source links at the top of the results page.
AI 리퍼럴 트래픽
Visits generated when users click source links within AI answers, including ChatGPT, Perplexity and AI Overviews.
AI 검색
Search that uses large language models to understand context and synthesize web results into natural-language answers rather than relying solely on keyword matching.
AI 검색 최적화
The work of making website content easier for generative AI search engines to understand and consider as a trustworthy source.
AI 슬롭
Low-quality content mass-produced indiscriminately with AI to pursue views or search visibility without delivering value to users.
AI 가시성
The frequency and share of brand appearances in answers from generative AI platforms such as ChatGPT, Claude and Perplexity.
답변 엔진
A system that provides complete summarized answers to complex questions rather than simply listing links to web documents.
AEO
A strategy for optimizing content so that brands and their content can be cited in the single answers generated by AI search systems or voice assistants.
브랜드 멘션
A textual reference to a brand or product on the web, whether or not it contains a hyperlink.
사고 사슬
A prompting technique that encourages intermediate reasoning steps to improve an LLM's accuracy on complex problems.
ChatGPT search
OpenAI's integration of live web search into ChatGPT, providing current web information and source links through a conversational interface.
청킹
Preprocessing that divides large source documents into meaningful pieces sized for retrieval and embedding in a RAG system.
Claude 웹 검색
A feature that enables Anthropic's Claude models to invoke live web search and reference current documents in their answers.
컨텍스트 엔지니어링
The comprehensive practice of selecting instructions, conversation history, retrieved documents and tools to create the most useful input for an AI task.
컨텍스트 윈도우
The maximum number of tokens a language model can consider within a single request-and-response interaction.
임베딩
A technique that represents the meaning of words, sentences or documents as coordinates in a multidimensional numerical vector space.
엔티티
An independently identifiable concept, such as a person, place, organization or object, with a distinct definition.
파인튜닝
Additional training that adjusts a pretrained model's weights using a dataset specific to a domain or task.
함수 호출
A capability that allows an LLM to identify when an external function or API is needed and generate its arguments in JSON.
Gemini Deep Research
A Google Gemini capability that investigates hundreds of web sources through multiple research stages and synthesizes them into a professional report.
GEO
A marketing strategy designed to improve the opportunity for a brand to be cited in answers from AI-powered generative engines as well as search engines.
그라운딩
The process of connecting AI answers to trustworthy factual material, such as verified web documents or databases, to reduce unsupported generation.
할루시네이션
A phenomenon in which an AI language model invents false information or sources and presents them as plausible facts.
하이브리드 검색
A retrieval approach that combines traditional keyword-based lexical search, such as BM25, with vector search based on semantic similarity.
지식 컷오프
The point in time through which a pretrained language model's training data was collected.
지식 그래프
A knowledge database that organizes entities and their relationships as a network of nodes and edges.
LLM
A deep-learning neural network pretrained on extensive text data to understand and generate language in a humanlike form.
LLMO
The work of helping different LLMs recognize a brand and its products as accurate, positively represented entities.
llms.txt
A Markdown-based proposal that helps LLMs and AI agents quickly identify a website's essential information and documentation.
마이크로소프트 코파일럿
Microsoft's AI assistant, combining Bing search with GPT-family models and integrated across Windows, Edge and Office applications.
MCP
A protocol that enables AI models to interact with external data sources and tools through a standardized, security-conscious interface.
멀티모달 검색
Search that understands and processes multiple input formats, including text, images, audio and video.
네이버 AI 브리핑
Naver's AI search service that presents concise answers and credible sources at the top of integrated search results.
네이버 스마트블록
Naver's integrated-search presentation that groups results for a query into topic-specific blocks such as frequently sought or well-received content.
Perplexity
An AI search service that combines live web retrieval with LLM-generated summaries and explicit source footnotes for each answer.
프롬프트
The instructions and contextual text supplied to an AI language model to elicit a desired answer or action.
프롬프트 엔지니어링
The practice of designing instructions, structure and examples to obtain better results from an AI model.
프롬프트 인젝션
An attack that uses malicious input to override an AI system's intended instructions or safety constraints.
쿼리 팬아웃
A technique that divides a complex question into multiple subqueries, searches them in parallel and synthesizes the results.
쿼리 재작성
The process of converting ambiguous or conversational questions into explicit queries that help a retrieval system find accurate results.
RLHF
A training approach that uses human preference judgments to build a reward model and align LLM outputs with human intent.
리랭커
A component that precisely reassesses the semantic relevance of initially retrieved documents with a cross-encoder model and reorders them.
RAG
A technique that retrieves current information from external documents or the web and adds it to an LLM prompt to support accurate answers.
시맨틱 캐싱
A performance and cost optimization technique that reuses earlier LLM answers for semantically similar questions based on embedding similarity, even when the wording differs.
Share of Model
A metric measuring how frequently AI models recommend or mention a brand relative to competitors for questions in a given industry.
토큰
The smallest unit of text processed by an AI model, which may be a word, character or subword fragment.
벡터 데이터베이스
A specialized database that stores numerical representations of content and searches them rapidly by semantic similarity.
음성 검색
Search conducted through spoken questions to a smart speaker or mobile voice assistant, typically returning a single spoken answer.
제로클릭 검색
A search completed without clicking an external website because information in a SERP summary or knowledge panel satisfies the user.