Principle 1: Write Clear and Specific Instructions

This principle focuses on guiding the AI with clarity so it understands exactly what is expected. Clear prompts reduce confusion and improve accuracy. One important tactic is using delimiters, such as triple quotes, brackets, or phrases like “Text:” to clearly separate instructions from the content. Another tactic is asking for structured output, for example requesting answers in bullet points, tables, or numbered steps, which makes responses organized and easy to use. Prompts should also state conditions and constraints, such as word limits, grade level, tone, or format, so the response matches expectations. Few-shot prompting is another useful tactic, where you provide examples of input and output to show the AI what a good answer looks like. Overall, being specific does not mean being complicated; it means being precise, focused, and intentional so the model delivers exactly what you need. Some tactics are :

This principle helps improve reasoning and accuracy by encouraging the AI to think step by step instead of rushing to an answer. A key tactic is asking the model to break the task into steps, such as analyzing the problem first, then solving it, and finally explaining the result. Another tactic is to explicitly request reasoning, for example by saying “explain your thinking” or “show the steps before the final answer.” This is especially useful for math, logic, or complex questions. You can also ask the model to work out its solution before giving the final response, which reduces mistakes and improves depth. For comparison or evaluation tasks, instructing the model to consider multiple options before choosing one leads to more balanced and thoughtful answers. Giving the model space to think turns quick guesses into well-reasoned solutions