System Prompt
A special instruction set given to an AI model before the user's message, defining the model's persona, behavior rules, output format, and constraints. The foundation of custom AI assistants and chatbots.
Why it matters
The system prompt is the hidden instruction that shapes how an AI behaves before you type anything. It sets the persona, the rules, the tone, and the format. Most custom assistants and "GPTs" are really just a good system prompt wrapped around a standard model. Learning to write clear ones is the single highest-leverage skill for getting consistent, on-brand results from any AI tool.
A concrete example
You want an assistant that always replies as a friendly fitness coach in three short bullet points and never gives medical advice. You put exactly that in the system prompt: the role, the format, and the boundary. From then on, every user message is answered in that style automatically, without the user having to repeat the instructions or even know they exist. Change the system prompt and the whole personality changes.
How to use it
Use it for the things that should be true of every response — role, scope, tone, what to do when it does not know — and keep the per-request prompt for the actual task. Being explicit about limits matters most: an instruction to say so and hand over when a question is out of scope prevents the single most common failure in customer-facing assistants.
The common mistake
Putting secrets or hard security rules in it. It is a strong default, not a boundary; users can often get it revealed or worked around, so anything that must not happen belongs enforced in code.
Related terms
Prompt
The text instruction you give to an AI model to generate a response. Prompt quality directly impacts output quality — better prompts yield dramatically better results.
Guardrails
Safety mechanisms built into AI systems to prevent harmful, biased, or off-topic outputs. Includes content filters, topic restrictions, output validation, and behavioral boundaries that keep AI responses within acceptable limits.
Prompt Caching
A performance optimization that stores and reuses the processed representation of repeated prompt prefixes. Reduces latency and cost when sending similar prompts — especially useful for system prompts and few-shot examples.
A/B Testing (for Prompts)
Running two or more prompt variations against the same AI model to compare output quality. Data-driven prompt optimization that consistently outperforms intuition-based prompting.
Chain-of-Thought (CoT)
A prompting technique that instructs the AI to reason step-by-step before giving a final answer. Dramatically improves accuracy on complex reasoning, math, and logic tasks.
Few-Shot Prompting
Providing 2-5 examples of desired input-output pairs in your prompt before asking the AI to perform the task. Significantly improves output quality for specialized tasks.