Lesson 14 of 1570
System Prompts vs User Prompts
Every AI conversation has two layers: a system prompt that sets the rules, and user prompts you type. Understanding the difference is the gateway to building AI-powered tools.
Lesson map
What this lesson covers
Learning path
The main moves in order
- 1Two channels, one conversation
- 2system prompt
- 3user prompt
- 4message roles
Concept cluster
Terms to connect while reading
Section 1
Two channels, one conversation
When you chat with an AI in a consumer app, you only see the user messages. But behind the scenes, most AI tools send a system prompt first. That prompt sets the AI's personality, rules, and job — and the user can't see it or overwrite it easily.
Compare the options
| System Prompt | User Prompt |
|---|---|
| Set once, applies to the whole conversation. | Sent fresh each message. |
| Defines role, rules, tone, limits. | Asks specific questions or tasks. |
| Usually invisible to the end user. | Visible in the chat. |
| Written by the developer. | Written by the person using the app. |
What a system prompt looks like
The system prompt sets the frame; the user prompt is the actual question.
[SYSTEM]
You are HomeworkHelper, a tutor for students ages 11-14.
Rules:
- Never give the final answer to a math problem. Always guide the student to find it.
- Keep responses under 120 words.
- If asked about anything off-topic (celebrities, dating, violence), politely redirect to schoolwork.
- Use encouraging, patient language. Celebrate small wins.
[USER]
I don't get how to factor x^2 + 5x + 6.That system prompt is the soul of the product. Change 'never give the final answer' to 'always give the final answer with no explanation' and you have a completely different tool — even though it's the same underlying model.
Layered control
- 1Model training — baked-in values and capabilities (Anthropic, OpenAI, etc.).
- 2System prompt — the product's rules and personality.
- 3User prompt — the immediate request.
- 4Follow-ups — refinements within the conversation.
Can you see the system prompt?
Sometimes. Some products publish theirs (Anthropic publishes Claude's Claude.ai system prompt). Others are secret. Users sometimes try to trick the AI into revealing its system prompt — a 'prompt leak.' Most modern AI tools are trained to resist this.
Key terms in this lesson
End-of-lesson quiz
Check what stuck
15 questions · Score saves to your progress.
Tutor
Curious about “System Prompts vs User Prompts”?
Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.
Progress saved locally in this browser. Sign in to sync across devices.
Related lessons
Keep going
Creators · 40 min
System Prompt Architecture: Design, Layering, and Policy, Part 2
When the system prompt and the user message disagree, design which one wins on purpose.
Builders · 26 min
Few-Shot Prompting: Teach by Example
Instead of describing what you want, show the AI two or three examples. Few-shot prompting is often the fastest way to get consistent output.
Builders · 24 min
Chain-of-Thought: Make the AI Show Its Work
Telling the AI to 'think step by step' before answering dramatically improves its accuracy on reasoning problems. Here's why and when.