Six principles for better AI prompts

The core communication framework for clear, specific, and useful responses — clarity, examples, structured reasoning, iteration, leverage, and role-playing.

Adapted from Anthropic · January 15, 2026
FoundationsBeginner8 min read

Great prompts are not magic formulas — they are clear communication. These six principles help teams get consistently useful outputs from any AI model.

1. Be clear and specific

State the task, audience, and desired format up front. Vague requests produce vague answers.

2. Use examples

Show one or two examples of the output you want. Examples anchor tone, structure, and level of detail.

3. Ask for structured reasoning

When accuracy matters, ask the model to think step by step before giving a final answer.

4. Iterate deliberately

Treat prompting as a conversation. Refine constraints based on what worked and what did not.

5. Leverage context

Provide relevant background, constraints, and definitions so the model does not guess.

6. Use role prompting

Assign a perspective — strategist, editor, analyst — to shape voice and priorities.