You don’t need to memorize every AI term before getting started, but a handful come up constantly enough in the first week that knowing them upfront saves real confusion.
Prompt. Simply what you type or say to the AI, your actual request. “Prompting well” means writing clear, specific requests, covered in our full prompt engineering guide.
Context. Everything the model can currently “see”: your prompt, attached files, the conversation so far. More context generally means better answers, up to a real limit.
Hallucination. When a model states something false with total confidence. Our full explainer covers why this happens, worth understanding early since it affects how much you should trust specific claims.
Agent. An AI system that takes multi-step actions toward a goal, rather than just answering a question. Token. The basic unit a model reads and generates. It shows up constantly in pricing, and our full explainer covers what it means in practice. Bookmark our full AI glossary for anything else that comes up, and see Anthropic’s own help center as one starting point.
