A prompt is the text you send to the model. It can be a single user message or a structured conversation (system message + user + assistant turns). Better prompts get better results: be specific about what you want, set a role or format (e.g. "You are a teacher. Answer in 3 bullet points."), and break complex tasks into clear steps. You don’t need to be a programmer — clear, concrete instructions and examples (few-shot) work best.
Tips for better prompts
- 1.Be specific: "Summarize in 3 bullet points" beats "Summarize".
- 2.Give role or format: "You are a teacher. Explain like I'm 10."
- 3.One task per prompt; break complex asks into steps.
- 4.Include examples (few-shot) when the output format matters.
- 5.If the answer is wrong, try rephrasing or adding constraints.
Example: Vague vs specific
Vague: "Tell me about Python." → The model may give a long, generic intro. Specific: "In 3 bullet points, list the main differences between Python 2 and 3 for a beginner." → You get a focused, scoped answer. Adding "You are a senior developer" can further steer tone and depth.