Prompt Engineering Best Practices

Basic Techniques

• ☐Provide examples (few-shot prompting) for complex tasks
  Show the AI what you want with 1-3 examples
• ☐Design with simplicity - start simple and add complexity
  Don't overcomplicate prompts upfront
• ☐Be specific about the output format you want
  JSON, bullet points, table, etc.
• ☐Use instructions over constraints where possible
  Tell the AI what to do, not just what not to do

Advanced Techniques

• ☐Use Chain of Thought (CoT) prompting for reasoning tasks
  Ask AI to 'think step by step'
• ☐Apply role prompting when domain expertise is needed
  e.g., 'You are an experienced marketing consultant'
• ☐Use system prompts to set consistent behavior
  Define the AI's personality and rules upfront
• ☐Try step-back prompting for complex problems
  Ask for general principles before specific answers

Quality Control

• ☐Control max token length for consistent output
  Set limits to avoid overly long responses
• ☐Experiment with temperature settings
  Lower for factual, higher for creative
• ☐Document your prompt iterations
  Keep track of what works and what doesn't