New

2,000+ fresh prompts added this week — grab the latest bundle for $47.99

  • Home
  • »
  • AI Guides
  • »
  • What a system prompt is, and when you need one

What a system prompt is, and when you need one

A system prompt is the standing instruction that applies to an entire conversation, rather than a single message. It is where you put the things that should stay true no matter what you ask next.

What belongs in it

Anything durable. Who the model is being. Your brand voice. Formatting rules. Things it must never do. Facts about your business that every answer depends on. If you find yourself pasting the same paragraph at the top of every message, that paragraph belongs in a system prompt.

What does not

The specific task. The material to work on. Anything that changes between messages. A system prompt stuffed with one-off details gets in the way, because the model keeps applying instructions that no longer apply.

When you do not need one

For a single question you will not repeat, it is overhead. System prompts pay off with repetition: a support workflow, a content pipeline, anything a team runs the same way every day.

A useful test

  • Will this instruction still be true in my next message? Then it is a system prompt.
  • Is it about this particular task? Then it belongs in the message.
  • Am I repeating it constantly? Move it up.

Most tools now let you save these as custom instructions, projects or personas. Whatever the label, the principle is the same: separate the standing rules from the request.

Other Guides

Why AI outputs sound generic, and how to fix it

Generic output is a predictable result of a generic prompt. Three changes that fix it in most cases.

Using AI safely with customer data

What to check before pasting anything that belongs to a customer into an AI tool.

Building a prompt library your team will actually use

Why most internal prompt libraries go stale, and what the ones that survive do differently.

When not to use AI

The tasks where reaching for a model costs more than it saves, and how to recognise them early.