Glossary

AI glossary

Short, practical definitions of the AI and marketing terms you will meet in real work.

AI search

GEO (Generative Engine Optimization): GEO is the practice of making content easy for generative search and answer engines to understand, trust and cite. The goal is not only to rank in a list of links, but to become a source inside an AI-generated answer.
AEO (Answer Engine Optimization): AEO structures content so search engines and AI systems can use it as a direct answer. Clear questions, concise answers and trustworthy sources are central.
LLMO (LLM Optimization): LLMO is another name for optimizing content so large language models can understand, mention and cite it in their answers.
AI Overviews: AI Overviews are Google's generated summaries shown above traditional search results. They answer the query directly and link to selected sources.

LLM (large language model): An LLM is a model trained on large amounts of text to predict and generate language. ChatGPT, Claude and Gemini are products built around language models.
Token: A token is a small unit of text processed by a language model. Context limits and API prices are commonly measured in tokens.
Context window: The context window is the maximum amount of information a model can consider at once, including prompts, documents and previous messages.
Hallucination: A hallucination is a plausible-sounding but false claim generated by AI. Important facts, numbers and citations require verification.
Multimodal AI: Multimodal AI can work across several forms of information, such as text, images, audio and video.
Open-weight model: An open-weight model makes its trained parameters available for download and local use. It is not necessarily fully open-source.
Fine-tuning: Fine-tuning adapts an existing model using examples from a specific task or style. It is more involved than prompting or source-grounded retrieval.

Prompt and prompting: A prompt is the instruction, context or question given to an AI model. Good prompts make the goal, context and expected output clear.
AI agent: An AI agent can carry out multi-step work, use tools and check progress toward a goal instead of only returning one answer.
AI skill: An AI skill packages a repeatable workflow, instructions, references and checks so an agent can perform a task consistently.
RAG (Retrieval-Augmented Generation): RAG retrieves relevant information from defined sources before generating an answer. It improves traceability and reduces unsupported claims.
MCP (Model Context Protocol): MCP is an open protocol that gives AI clients a standard way to connect to tools and data sources.
Embedding: An embedding converts content into a numerical representation of meaning. It is widely used for semantic search and retrieval.

MMM (Marketing Mix Modeling): MMM estimates how marketing channels contribute to business outcomes without relying on individual user tracking.
Attribution: Attribution assigns credit for a conversion across marketing touchpoints. Different models produce different views of channel value.