Do you use an LLM such as ChatGPT, Gemini, or Claude on a daily basis, but find that the results are sometimes disappointing ? Do you have to rephrase the same request three times before getting something usable ? More often than not, the problem isn’t with the AI, but with your prompt, which directly determines the quality of your result. Knowing how to talk to AI is becoming a key skill in business.
RICARE: the method that transforms your rough prompts into laser-like instructions
ADN IA has developed RICARE, a six-point method that enables you to formulate clear, precise, and effective requests.
R as ROLE
Start there. Defining the role of AI means shining a spotlight on the area of expertise you want to draw on, among the hundreds of billions of parameters in its model. A prompt without a role is like asking someone for advice without knowing if they are familiar with the subject. Example:
“You are an expert in marketing and copywriting, capable of writing compelling communications.”
I as INSTRUCTIONS
Your instructions should be clear, concise, and affirmative (an action verb in the imperative). Example: “Write an analysis plan, summarize this text, improve this content, create a graph...”
Structured by separators (###Example 1###) if your instructions contain documents to be processed, examples to follow, templates to customize, etc.
If the request is complex, divide it into several separate prompts, each handling one step of the task.
Pro tip: Not sure how to phrase your request, or worried you might forget something? Ask the AI some preliminary questions to help it build the optimal prompt itself.
C as CONTEXT
It transforms a generic response into a tailor-made solution.
- Provide concrete examples.
- Specify the business context: “An SME in the public works sector.”
- Specify the expected characteristics, use cases, etc.
- Add restrictions if necessary: “answer within the framework of French law only.”
A telling example:
- ❌ Bad: “Determine the sentiment of this post [INSERT POST]”
- ✅ Good: “Determine the sentiment of this post using these examples: ‘great pen’ → positive / ‘I hate it when my battery dies’ → negative [INSERT POST]”
A as AUDIENCE
Who will read or use this response? How should we address them? The AI needs to know this.
Explain quantum computing to me as if I were 10 years old. Use simple, educational language, examples, and illustrations.
Or even:
“Your report is intended for members of the Executive Committee who are expecting strategic recommendations and a focus on growth opportunities. Use a precise style and a cordial but professional tone.”
Adapting the tone, style, and complexity to your audience ensures that your content will hit the mark.
R as RESPONSE FORMAT
Be explicit about the expected format. AI cannot read your mind.
- Format : table, bullet points, list, tweet, HTML code…
- Length restrictions: less than 50 words, 10 points, 300 characters, 1 page…
- Language and syntax: French, formatted text, markdown, JavaScript code, etc.
- Séparateurs spécifiques : tags, points-virgules, colonnes…
The more specific you are, the more satisfied you will be with the answer.
E as EVALUATION
The perfect prompt doesn’t exist on the first try. Excellence comes from iteration.
Two key moments for evaluation:
- In your initial prompt, validate the AI’s understanding: “Do you understand? Feel free to ask me questions to clarify your mission and improve your response.” You will see that it can broaden your analysis through its questions.
- Feedback on the response: Give feedback, adjust, refine. Clicking on the thumbs-up icon means improving the model a little for next time. Giving feedback means iterating to improve the result. “Great! Add more examples like this, but in a less formal style. And be more concise.”
Evaluation transforms a good result into an exceptional result.
The difference between an amateur and an expert? The method.
Because learning to prompt is like learning a new language. A language that opens the door to a powerful assistant, available 24/7, capable of revolutionizing your productivity. And that requires mastery of the method and its practical application.
Take action now : join the ADN IA training
Want to move from the “I’m trying” level to the “I’ve mastered it” level?
The ADN IA training course in prompt engineering gives you the keys to:
- Structure your prompts like an expert from the very first attempt
- Unlock the full potential of generative AI in your business
- Save hours each week by automating intelligently
- Create AI workflows that deliver professional results
AI will not replace your expertise. But those who master AI will replace those who do not.
Ready to get a head start?👉 Discover the ADN IA training courses and transform the way you work with artificial intelligence.