Every chat with a model has two layers. The system prompt is set once, before the conversation starts, and defines who the model is, what rules it follows, and how it should respond. The user prompt is the specific request you send each turn. Putting standing instructions in the right layer makes responses more consistent and saves you from repeating yourself.
The lessons cover what a system prompt does and how to write one, how it differs from the user prompt, and a closer look at role prompting: when telling a model to "act as an expert" shapes tone and style, and when it does little for accuracy. By the end you should know what belongs in a system prompt and what should stay in the request.
Watch the lessons in order, then continue on to Examples & Few-Shot Prompting.
A clear introduction to the instruction layer that runs before every message: how it differs from the user prompt, why models follow it more strictly, and a worked example built from a one-line role into a full set of rules.
Video 8minA deeper look at system prompts as a foundation of modern AI systems: what they control, the techniques used to write them, and the problems that show up when instructions conflict or get ignored.
Video 15minDan Cleary reviews the research on personas: they clearly change tone and style, but add little for accuracy-based tasks, and a specific, detailed role works better than a one-liner.
Video 14minA practical look at using system prompts to stop re-running prompts and hoping, and start building repeatable workflows that behave the same way every time.
Video 6min