Prompt Engineering: The PAPA-FTR Framework
PAPA-FTR is an AI prompting framework created by Jeff St Louis that teaches AI users how to effectively write prompts that help them get the best out of artificial intelligence.
What Does PAPA-FTR Stand For?
PAPA-FTR stands for Persona, Audience, Purpose, Application, Format, Tone, and Reference. Below is a brief presentation of this framework.
Persona
The expert role or identity you want the AI to take on. It frames the AI’s reasoning, voice, and perspective.
Ex: “You are an HR expert with 25 years of experience.”
Audience
This is the intended readers, target audience, or recipients of the AI’s output, along with any relevant information about them.
Ex: “The recipients are my company’s salesmen, each of whom comes from a different cultural background.”
Purpose
The reason or goal for creating the content. This information puts you and the AI on the same page.
Ex: “Inviting qualified candidates to apply for the social media marketing position.”
Application
The platform or context where the output will be used. It helps with the length and structure of the answer.
Ex: “This will appear on the company’s careers page and LinkedIn business page.”
Format
The desired layout or presentation style of the output: do you want bullet points or paragraphs?
Ex: “Organize your answer in sections and sub-sections.”
Tone
The voice, mood, or personality of the writing. Think of your target audience when choosing the tone.
Ex: “Use a professional and inviting tone.”
Reference
Any source, style guide, or sample that the AI should follow. (Optional)
Ex: “Use the attached PDF document as a guide for the presentation of your answer.” (Avoid using more than 2 references for better a performance.)
This framework will put you and the AI chatbot on the same page, but it doesn’t necessarily guarantee the perfect answer you‘re looking for. You may still need to evaluate the AI’s response to see if you missed anything in your input or if the AI chatbot either missed or misinterpreted anything from your prompt. After evaluating the output, you may need to iterate until you get the perfect answer you’re looking for. It may take one prompt or more to get what you want, depending on the AI model you’re using.