Basics of Prompt Engineering

Structure you prompt

An effective prompt should be carefully structured to obtain the best results. You don’t necessarily need to use all the following areas, but an ideal prompt should be structured something as below:

Context and/or persona

Why you need the request and/or who is needing it

Goal

What task you want to achieve

Source

Where to find the data

Expectations

How you want the output to be structured

Even if not all areas are always mandatory, consider adapting your prompt to provide more details if the output is not satisfying.

Structure examples

Good practices

Consider these best practices to obtain the best results when using Generative AI.

1. Always be clear and concise

   Don’t bother yourself in complex sentences. Always focus on only providing the details that will be relevant to your request.

2. Use natural language

   Express yourself completely, in full sentences, just like if you were talking with someone else.

3. Iterate and refine

   You’ll not always obtain the best answer directly. Iterate and refine in the same conversation, by asking AI to precise or provide more context until you obtain the desired outcome.

4. Review the output

   GenAI can make mistakes. Always remember to review and check the outcome before using it.

Back to top