Structured Prompting
Move beyond 'just asking' to structured prompts that produce consistent, high-quality results every time. XML tags, chain-of-thought reasoning, and multi-step prompts turn AI from a toy into a tool.
Use XML tags to separate sections of your prompt: <context>, <task>, <format>, <examples>. This helps the AI parse complex instructions.
Chain-of-thought: ask the AI to 'think step by step' before giving its final answer. This dramatically improves accuracy on complex problems.
For repeatable tasks (weekly reports, email templates), save your best prompt as a template. Swap in the variable details each time.
If the AI gets something wrong, don't rewrite the whole prompt. Add a specific instruction addressing the error: 'Do NOT include X' or 'Make sure to Y'.
TEMPLATE STRUCTURE: Role → Context → Task → Format → Constraints → Examples
Wrap distinct sections in XML tags for clarity
Use <thinking> tags to request step-by-step reasoning before the final output
For complex tasks, break them into numbered sub-tasks
Provide POSITIVE examples (what to do) AND NEGATIVE examples (what to avoid)
Test your prompt with different inputs to check it generalises well
Build a personal prompt library — save templates for recurring tasks
Use system prompts (in the API) to set persistent behaviour across a conversation