Glossary
AI glossary
The most-used artificial intelligence terms, explained simply and with examples for agencies. From MCP and RAG to tokens, agents and prompts. Choosing between tools? See the alternatives comparisons or the bots you set up with a single prompt.
Summary
This glossary is a collection of 20 artificial intelligence terms, each with a practical example for agencies. Every definition is written for people who build with GoHighLevel, Claude Code or Codex.
Updated
AI agent
An AI agent is a system that receives a goal, decides which steps to take, uses real tools and acts until it gets there, with little supervision.
API
An API is an interface that lets two programs communicate: one asks for something and the other responds, following defined rules.
Context window
The context window is the maximum number of tokens a model can consider in a single interaction, input and output combined.
Embeddings
Embeddings are numeric representations (vectors) of text that capture its meaning, so AI can compare and search by similarity.
Few-shot
Few-shot is the technique of including in the prompt a few examples of the result you want, so the model imitates that pattern.
Fine-tuning
Fine-tuning is retraining a base model on your own examples so it adopts a constant style, format or behavior.
GEO
GEO (Generative Engine Optimization) is optimizing your content so AI engines (ChatGPT, Perplexity, Google AI Overviews) cite it in their answers.
Hallucination
A hallucination is when a model generates false or made-up information but presents it with confidence as if it were true.
LLM
An LLM (Large Language Model) is a model trained on huge amounts of text that predicts and generates language, the brain behind ChatGPT, Claude or Gemini.
MCP
MCP (Model Context Protocol) is the open standard that gives an AI agent real tools: search, read data or control apps through natural language.
Prompt
A prompt is the instruction or input you give an AI model to guide its answer. Its quality sets the quality of the result.
Prompt engineering
Prompt engineering is the practice of designing and refining prompts (role, context, format, examples) so a model delivers better, more consistent results.
RAG
RAG (Retrieval-Augmented Generation) is the technique of giving a model, at answer time, the relevant pieces of your documents so it responds with your information.
Subagent
A subagent is a specialized agent, with its own context and instructions, that a main agent calls to delegate a bounded task.
System prompt
The system prompt is the base instruction that sets the model's role, tone and rules for the whole conversation, above each user message.
Temperature
Temperature is a parameter that controls the randomness of answers: low makes them more predictable and consistent, high makes them more creative and varied.
Token
A token is the smallest unit of text a model processes (roughly a short word or part of one). Usage and cost are measured in tokens.
Vector database
A vector database stores embeddings and finds the ones most similar to a query, enabling search by meaning.
Webhook
A webhook is an automatic message one system sends to a URL of another system the moment an event happens.
Workflow
A workflow is an automatic sequence of steps (trigger, conditions, actions) that runs on its own when an event happens.