The fine art of asking the right AI question

Home News The fine art of asking the right AI question

Prompting is one of the most important, yet most underrated, skills when working with generative AI. A prompt is what you tell AI to do. Think of it as the first note of a song: it sets the tone for everything that follows. After Tara Kikken and I, Jéssica Perozo, taught several workshops on AI prompting to Valid colleagues, we heard the same response over and over again, "If only I had known this earlier." Regardless of which AI tool you use, these techniques are widely applicable. In this blog, I'll show you why prompting matters and how to get it right.

Why generative AI cannot do without a good prompt

The result of AI often disappoints: it doesn't feel right, comes across as biased, is vague or simply irrelevant. Although AI is developing rapidly and these limitations are getting smaller by the day, it is still important to know how to manage AI effectively to get the best and most reliable result.

AI is predictive - not smart

Most AI tools are based on a Large Language Model (LLM): a system trained on massive amounts of text from the Internet. Such a model uses a neural network (a smart algorithm inspired by how the human brain works) to recognize patterns, relationships and context in language and make predictions about which words logically follow your input.

It is important to note that AI does not think or understand like a human, but is exceptionally good at making language predictions. In short: if your input is vague or biased, so is your output.

"Why do I get a different answer every time?"

A common question asked during the workshops I give on AI is, "Why do I get a different answer every time?" The answer: temperature. Many AI models use a setting called temperature, which determines the degree of randomness.

Low temperature (e.g., 0.2) = safer, more consistent and often more boring.
High temperature (e.g., 0.8) = more creative, but less predictable.

That's why sometimes you get a different answer with the exact same prompt.

Common AI limitations:

Complexe of overvolle prompts en hallicunaties

The 5 elements of a strong prompt

Writing a good prompt is not about expensive words, but about clarity, structure and intent. Here's a simple formula that works for any AI tool.

Instruction - What should AI do?
Context - What background information is needed?
Question - What exactly do you want to know?
Input data - What information should be used?
Output format - How should the answer be delivered?

Bad prompt: "Help me with my presentation." AI doesn't know what about, for whom, or what help you need.

Strong prompt:

Instruction: Write a short presentation.
Context: I present to management on project results.
Q: What are the 3 main conclusions?
Input: The project reduced onboarding from 6 to 3 weeks.
Output: 3 bullet points in business language suitable for slides.

Reply from AI:

Onboarding time cut in half (from 6 to 3 weeks)
90% positive feedback from new employees
€120K estimated annual cost savings

Tips & tricks to improve your prompting

Here are some helpful tips & tricks to get the most out of prompting:

Contextual preparation (prompt-persona)
Give AI a role or character:
"Pretend you're Steve Jobs. Give feedback on my business plan."
Reverse prompts
Let AI write the prompt for you. Use the 5-element structure and end with:
"Write a strong prompt to achieve this goal."
Step-by-step reasoning via prompting
Add to your output request:
"Perform a chain of thought analysis."
This helps AI reason step-by-step.
Review & improve
Let AI review your prompt or document with a direct instruction:
"Critique and improve.
Use direct language
Avoid cautious wording such as "Can you..." or "Please..."
Use action-oriented instructions for clearer results:
"Summarize this in 5 bullet points."

One final prompting principle

This is how I personally decide whether to engage AI:
"Am I taking way too long with this?"
If so: time to bring in AI.

Note: If what you want to enter makes your heart race at the idea of it appearing on the news ... then better not share it.

Summary

Prompting is not just typing something and hoping for the best; it is a profession. The better your prompt, the smarter your AI and the more fun and useful it becomes. So the next time you stare at that blinking cursor: don't just ask a question, prompt with intent. Structure. Shape. Direct. Because in the world of generative AI, your words are the code. Your words are the brush, AI is the paint. Make something beautiful out of it!

Get in touch

About Jéssica Perozo

This blog was written by Jéssica Perozo, consultant at Valid. From her background as BIA and PMO, she helps organizations to design processes smarter and future-proof. Always with a focus on adoption by people. With methodologies such as Lean Six Sigma, Design Thinking and Change Management, she is constantly looking for ways to improve work: by automating, standardizing and above all simplifying. Whether it involves AI, training or change projects, Jéssica combines analysis with a practical, people-oriented approach. This is how she realizes improvements with lasting impact.