Glossary · Technical
System Prompt
Hidden instructions set by the AI platform itself — not the end user — that shape a model's tone, formatting, and sometimes which sources it prefers or avoids.
What it is
Before a user's question ever reaches the model, the platform prepends its own instructions: how to format answers, whether to cite sources, how to handle sensitive topics, sometimes even which categories of content to avoid recommending. Users never see this text, but it applies to every conversation on that platform.
Why it matters for AI visibility
A system prompt that instructs a model to always cite sources produces very different visibility dynamics than one that doesn't — the first rewards being a citable, well-structured source; the second rewards simply being well-known enough to be recalled from training. Brands can't see or edit this layer, but it silently shapes how mentions happen on each platform.
See more terms in the full glossary, or read the AI Visibility 101 guide for the full picture.