Effective Prompt Engineering: How to Get Better Results from AI Models

TutorialsBy AI Nav Editorial Team5 min read

Why do two people typing into the same AI model receive vastly different results? Mastering prompt engineering is key to squeezing maximum performance out of LLMs.

The CRISPE Framework

A well-structured prompt usually includes these components:

  1. Capacity & Role: Assign a clear persona or domain expert role to the AI.
  2. Request: Clearly define the exact task or objective.
  3. Insight & Context: Provide background information, target audience, and constraints.
  4. Statement & Style: Specify output tone, formatting rules (Markdown, JSON, tables).
  5. Examples (Few-Shot Prompting): Supply 1-2 sample inputs and expected outputs for reference.

Practical Tips:

  • Use Delimiters: Separate context from instructions using triple backticks or markdown headings.
  • Chain of Thought: Prompt the model to "think step by step" to improve logical reasoning.
  • Negative Constraints: Explicitly specify what not to include.

Applying these frameworks ensures consistently high-quality outputs across all AI platforms.

Back to Blog