AEO Glossary.
Establishing a common vocabulary for the post-search era.
The technical process of optimizing content to be the single source of truth for Large Language Models (LLMs) like GPT-4, Claude, and Gemini.
The architecture used by AI engines (like Perplexity or Bing Chat) to fetch real-time data from the web before generating an answer. AEO targets this retrieval step.
When an AI confidently generates false information. AEO reduces hallucination risk by providing structured, verified data that the model can "ground" its answer in.
A footnote or direct link provided by an answer engine crediting the source of its information. This is the new "Click-Through Rate" metric.
A user query that is answered directly on the results page (or by the chatbot) without the user needing to visit a website.
A structured database of entities (people, places, things) and their relationships. Google and Bing use this to understand the "world" beyond just keywords.
A new paradigm where AI agents autonomously execute commercial transactions (browsing, selecting, purchasing) on behalf of users, utilizing protocols like MCP.
A probabilistic AI model trained on vast datasets to generate human-like text. Key examples include GPT-4, Claude 3.5 Sonnet, and Gemini Ultra.
The mathematical representation of text as numbers (vectors) in a multi-dimensional space, allowing AI to understand semantic relationships and context beyond exact keyword matches.
The limit of text an LLM can "remember" or process in a single conversation (e.g., 128k tokens). AEO aims to fit high-value content efficiently within this window.
A metric measuring how strongly an AI's response is supported by retrieved facts. High groundedness equals high trust and low hallucination.
A concept (patented by Google) describing content that provides *new* value or data unique to the entire dataset, rather than just restating existing knowledge.
The process of breaking text down into smaller units (tokens) for AI processing. AEO optimizes content density per token to maximize information retrieval.
An AI model's ability to complete a task correctly without having seen a specific example of that task during its training.
A synonym for AEO, focusing on optimizing content specifically for generative AI summaries rather than traditional blue search links.
Distinct people, places, organizations, or concepts that search engines recognize as unique objects in a Knowledge Graph, distinct from just strings of text.