System prompt: A system prompt is a set of instructions given to an AI model before a conversation starts that defines its role, tone, rules, and boundaries. Vendors write one for their products; you can add your own in custom instructions, projects, and custom assistants.
Every chat you have with an assistant begins with text you never see. The vendor's system prompt tells the model what product it is, how to behave, and what to refuse. Your messages are layered on top of that. The system prompt usually carries more weight than a normal message, which is why it is the right place for rules you want followed every time.
Most tools now let you write your own. ChatGPT has custom instructions and custom GPTs, Claude has project instructions, Gemini has Gems, and Microsoft Copilot has agents you can configure. Whatever the name, the idea is the same: write once who the model is, what it knows about you, and how it should respond, and every new chat inherits it.
Good system prompts are specific and short enough to maintain. Name the role, the audience, the house style, the formats you want, and the hard rules (never invent figures, always cite the source document, ask before assuming). Avoid stuffing in everything you might ever need; that dilutes the instructions that matter.
Example at work
A small accounting firm creates a shared assistant with this system prompt: 'You are a staff accountant's drafting assistant at a firm serving small businesses. Write in plain English for non-accountant owners. Never give tax advice; describe options and say a CPA will confirm. Format client emails with a one-line summary, then details, then next steps.'
Why it matters
A system prompt is the difference between re-explaining your job every morning and having an assistant that already knows it. It is also where a team encodes its rules, so output is consistent no matter who on the team is typing.
Related terms
- PromptA prompt is the text (and sometimes files or images) you give an AI model to tell it what you want. A good prompt states the role the AI should play, the task, the relevant context, the output format, and any constraints.
- Prompt engineeringPrompt engineering is the practice of writing and refining instructions to get reliable, high-quality output from an AI model. It covers giving context, showing examples, specifying format, and iterating on what did not work.
- Custom GPTA custom GPT is a tailored version of ChatGPT that you configure with your own instructions, uploaded reference files, and optional connections to outside services, then reuse or share without rewriting the setup each time.
- GuardrailsGuardrails are the rules, filters, and technical limits placed around an AI system to keep its behavior within acceptable bounds, such as refusing harmful requests, staying on topic, protecting data, or requiring approval before acting.
- ChatbotA chatbot is software you converse with by typing or speaking. Older chatbots followed scripted decision trees; modern AI chatbots such as ChatGPT, Claude, Gemini, and Copilot use large language models to understand and respond to almost any request.