What Is a System Prompt?
A system prompt is a set of instructions given to an AI model before a conversation starts, defining its role, tone, and rules for that specific use — separate from whatever the user actually types. It's the reason the same underlying model can act like a customer support agent in one app and a coding assistant in another.
Key takeaways
- A system prompt sets the AI's role, tone, and constraints before any user message arrives — it's configuration, not conversation.
- The same AI model behaves differently across different apps almost entirely because of different system prompts, not a different model.
- A well-written system prompt is specific about role, format, and what to avoid — a vague one produces inconsistent behavior across conversations.
- Most consumer chat interfaces hide the system prompt from the user entirely — it's set by whoever built the app, not typed by the person chatting.
Definition
A system prompt is a set of instructions provided to an AI model before a conversation begins, defining the role it should play, the tone to use, and any rules or constraints for that specific application — separate from and prior to anything the actual user types.
System prompt vs. user message
The system prompt is set once, by whoever builds the application, and applies to the entire conversation. The user message is whatever the person chatting actually types, one exchange at a time. A support chatbot's system prompt might say “only answer questions about our return policy and product catalog” — the user's message is whatever question they ask within that boundary.
A real example
A business deploys a chatbot with a system prompt instructing it to act as a helpful support agent, stay only within the business's real policies, and hand off anything unusual to a human. A customer's actual question (“can I return this after 30 days?”) is the user message — the system prompt is what keeps the bot on-topic and accurate across every different question it gets asked.
Why it matters for building with AI
If you're building anything on top of an AI model — a chatbot, an internal tool, an automation — the system prompt is where most of the actual behavior gets defined. A vague system prompt produces inconsistent behavior across different conversations; a specific one (clear role, format, and constraints) produces reliable, repeatable behavior.
Learn more
For the broader technique of writing effective prompts, see What Is Prompt Engineering? and the AI Glossary. AIVX Labs' Chatbot Builder tool lets members write and customize a real system prompt for their own deployed bot — see the AI Tools Directory.
Every course and tool mentioned here is included free on AIVX Labs.
Start freeFrequently asked questions
Can I see a chatbot's system prompt?
Usually not directly — most consumer apps don't expose it, since it's set by the app's builder, not the person chatting. Some AI platforms let developers write and see their own system prompts when building on top of a model.
Is a system prompt the same as a regular prompt?
No. A regular (user) prompt is the specific question or task typed into a conversation. A system prompt is set beforehand and applies to the whole conversation — it's configuration for how the AI should behave, not a single request.
Why do two apps using the same AI model behave so differently?
Almost always because of different system prompts — one app might instruct the model to act as a strict customer support agent staying only on-topic, while another instructs it to be a flexible general assistant. Same model, different configuration.