Glossary · Fundamentals
Training Data
The text an AI model learned from before it was released — the source of what it "knows" without retrieving anything new.
What shapes it
A model's training data is drawn from a large cross-section of the public web (and other licensed sources), captured up to some cutoff date. How often, and how consistently, a brand appeared across that data shapes what the model says about it by default.
Its limits
Training data has a cutoff and isn't updated in real time — which is part of why engines that retrieve live information (see: RAG, grounding) can reflect recent changes that a model's built-in knowledge alone would miss.
See more terms in the full glossary, or read the AI Visibility 101 guide for the full picture.