Prompting: Meaning, Examples, and Tips for Better AI Prompts

Article by: Job Jimmink

Last update: May 4, 2026

Prompting is the way you give instructions to AI. A prompt can be a question, a task, a short command or a longer briefing. The quality of that instruction has a clear effect on the answer you receive.

Many weak AI answers start with a prompt that is too broad or unclear. The model may still give an answer, but that answer can easily become too general, too long or not useful enough for the situation. A better prompt gives more context, explains the goal and makes clear what kind of output is needed.

In this article, you will read what prompting is and how it is used when working with AI. You will also learn how the TCREI model gives structure to a prompt, how prompting differs from prompt engineering and which mistakes often lead to weaker output. The article also includes examples for study, work and everyday tasks.

What is prompting?

Prompting means giving AI an instruction. The text you provide to an AI tool functions as a prompt. The AI then uses that prompt to create an answer.

A prompt can be short. For example: “write something about technology”. The project requirements appear straightforward but they create multiple unknown variables. AI does not know what kind of technology you mean, who the text is for or how detailed the answer should be.

When you add more context, the answer usually becomes better. You can explain the goal, the target audience and the type of output you want. A summary, list, table or step by step plan will each give a different result.

You can compare prompting with giving an assignment to a new colleague. The person would need to understand your requirements because you only stated “write a text about AI” without any additional details. You will obtain better results when you specify the goal and the target audience and the preferred output format. The result becomes more user-friendly when you specify the target audience and the desired format and explain the goal of the content.

A good prompt often contains three parts: context, instruction and goal. These parts help AI understand what you want to achieve. The answer will develop into an unsuitable solution which produces either an insufficient or an excessive or an unrelated output without these elements.

Background and origins

Prompting became widely known with the rise of AI tools such as ChatGPT. The core principles of prompting have existed since long before modern times. AI research has analyzed language models to determine how different instructions affect their generated responses. The outcome quality depends on both the model selection and the quality of input data according to evaluation results. The result of this discovery led to an immediately applicable outcome. The output will change significantly when you modify the input prompt by a small amount. The same AI model produces different results which include explanations and summaries and comparison and rewriting and idea generation based on the task requirements. The process of prompting has become essential to operating systems. The practical nature of this ability controls how well AI systems perform during their actual system deployment.

As AI tools became more widely used, prompting also became closely linked to prompt engineering. The process of prompting involves creating instructions which people use to perform specific tasks. Prompt engineering goes further and focuses on designing, testing, and refining prompts in a more systematic way, often for repeated use, workflows, or applications.

The research created new perspectives about how AI systems need to operate. Complex tasks achieve better outcomes when you separate them into individual processing stages. A single broad question often creates vague output. A structured prompt with clear direction usually leads to a more logical and more useful answer.

How do you use prompting?

Start your prompting process by identifying the exact information you want to obtain from AI systems. The response will become general when the requirement for the assignment expands to include various different subjects. That is why it helps to write a prompt as if you were giving someone a clear assignment.

People need to define AI system roles before they proceed with any practical implementation of AI technology. The model responds as different roles which include teacher and analyst and marketer and consultant and editor. The system provides specific instructions to the answer which leads to improved results that match the context more accurately.

The task needs to receive all necessary information which should be presented in an organized format. What type of response do you need between explanation and summary and analysis and rewrite and idea generation? The more precisely this is described, the easier it becomes for AI to stay on track.

Next, explain what kind of output you want. You can ask for a short paragraph, a table, a list of recommendations, or a step by step plan. The answer should include information about the intended readers and the appropriate communication style and the necessary amount of information and any specific requirements which must be followed.

The optimal solution for prompting requires a step by step approach during actual implementation. Start with a clear first version, review the output critically, and then improve the prompt where needed. The process generates answers which better match the intended purpose and suit the environmental conditions and follow the designated structure.

The TCREI model

For larger or more complex tasks, a loose prompt is often not enough. The answer can become too general, too long or just not specific enough.

The TCREI model helps you build a prompt step by step. It gives structure before you ask the question. That is useful when you want an AI tool to analyse something, write in a certain style, give advice or combine several requirements.

Task: start with the main task. What should the AI do? Do you want an analysis, summary, rewrite, checklist or recommendation? A clear task keeps the answer focused.
Context: add the background that matters. Think of the goal, target audience, situation, language, tone or limits. Without context, the answer may be technically correct, but still not useful for your situation.
References: give examples or source material. This can be a text, format, style example, article, dataset or earlier output. References help the AI understand what the result should look like.
Evaluate: check the answer critically. Is it correct? Is it complete? Does it match the task? Are there gaps, weak arguments or parts that sound too generic?
Iterate: improve the prompt and try again. Often one extra instruction makes a big difference. For example, ask for shorter sentences, fewer repeated points, more examples or a sharper conclusion.

The strength of TCREI is that it slows the process down just enough. You do not only ask for an answer. You give the AI the information it needs to produce a result that is clearer, more relevant and easier to use.

Common mistakes in prompting

Prompting often goes wrong because the instruction is too broad, too vague, or too unfocused. Users make a recurrent error when they attempt to obtain all their needed information during a single request. AI systems encounter difficulties when they must determine which components hold the most vital value. As a result, the answer may become fragmented, incomplete, or less useful.

Another common issue is missing context. AI needs to solve multiple unknown elements when users fail to specify their goals and their target viewers and their particular situation. The system produces routine answers which fail to address the specific requirements of the actual work task.

Users frequently do not establish their preferred format. The output will need substantial editing because you failed to specify your preference for summary or table or explanation or action plan. The procedure experiences delays which make the generated output less beneficial for practical applications.

The last error occurs when people accept the initial response without conducting any verification. AI systems can produce convincing output which contains both incorrect information and missing details while producing irrelevant results for the assigned task. The effectiveness of prompting depends on having straightforward instructions together with thorough evaluation methods.

Weak prompt vs. strong prompt

Strong prompts offer detailed instructions which weak prompts do not provide to the same extent. The weak prompt leaves all options completely unstructured. AI systems need to create their own requirements for target goals and intended viewers and output structure because of this situation.

A simple weak prompt is: “Write something about customer satisfaction.”. The prompt fails to provide any details about what the task requires or its intended use or its full scope. The response continues to provide general information which lacks detailed answers because of this situation.

A stronger version could be: “You are a marketing consultant. Write a short explanation of up to 200 words about customer satisfaction for aspiring entrepreneurs. Explain why customer satisfaction matters, describe three practical ways to improve it, and use an accessible and motivational tone.”

The new version functions better because it establishes the role and target audience and project boundaries and communication style and output requirements. The added direction leads to an answer which becomes more specific and immediately useful for the task at hand.

This example shows why prompting matters in practice. The creation of a better prompt does not ensure perfect answers will be received but it helps users to understand the task better while reducing their need for editing their work after submission.

Example of prompting

A student needs to create a report about a company which already exists. The assignment requires students to perform a DESTEP analysis which they can find at the following link: DESTEP. The student understands the model but seeks to apply AI technology which will produce an improved initial version at an accelerated pace.

The first prompt is simple: “Create a DESTEP analysis for Company A.”

The result will probably be too broad. The answer may mention trends that apply to many companies, but not clearly to this specific company. Some points may also be too general for the assignment. The AI tool functions correctly so it does not create any problems. The prompt gives too little direction.

A better prompt uses the TCREI model. The student explains the task, adds context and gives clear requirements for the output. For example:

“You are a business analyst. Create a DESTEP analysis for Company A in the [sector]. Focus only on external factors that are relevant to this company. Describe 2 to 3 factors for each DESTEP category. Use 100 to 300 words per factor. Use reliable Dutch sources, such as recent news articles. Take the grading rubric into account. Cite the sources and use relevant insights from the appendices.”

Researchers can obtain more useful information from this prompt. The task is clearer. The scope is smaller. The expected output is easier to check. The student should still review the answer carefully. The sources demonstrate their ability to be trusted. The factors need to prove their suitability for the company. The points contain sufficient details to fulfill the assignment requirements. The report lacks vital information which needs to be added to the document.

That final check matters. The process of prompting involves more than just creating superior questions because it requires users to study system responses for optimizing their prompts which leads to better outcomes.

Tips for writing better prompts

Prompting extends its usefulness to multiple situations which go beyond academic research activities. A marketer can use AI to draft campaign variations for different audiences. An HR professional can create a first version of a job posting with the right tone and requirements. A consultant can use AI to summarize interviews or structure long notes. In these situations, a stronger prompt often leads to faster output that needs less rewriting.

Better prompts usually start before you open the AI tool. First decide what the result should achieve, who it is for, and what form it should take. Then write the instruction as clearly as possible. This reduces the chance of generic output and makes the response easier to use.

AI needs to receive a well-defined purpose which enables it to execute its tasks effectively. Think of a teacher, analyst, editor, advisor, or marketer. Then describe the task in direct language. Explain what should be delivered and what the answer should look like.

At the same time, a good prompt does not remove the need for judgment. AI systems fail to detect contextual information while they extract information from untrustworthy sources which produces untrustworthy information that appears as reliable content. The combination of specific task directions with thorough feedback produces the most effective results through prompting.

Conclusion

Prompting is a practical skill for working with AI. The better the instruction, the easier it becomes for the AI tool to give an answer that fits the task.

A useful prompt does not have to be complicated. It should explain what needs to be done, why it matters, who the answer is for and what the output should look like. For larger tasks, the TCREI model helps to add structure before the AI tool starts generating an answer.

The final step is always human review. AI output can sound convincing, but still miss context, contain mistakes or need sharper examples. Good prompting therefore combines clear instructions with careful evaluation.

Frequently Asked Questions about Prompting

Prompting can improve AI output, but it does not solve every limitation of AI. The questions below add practical context that helps you use prompting more critically and more effectively.

When is prompting less effective?

Prompting is less effective when the task is unclear or when the answer depends on missing facts, recent developments, or human judgment. In these cases, a better prompt may still not produce a reliable result.

Why do similar prompts produce different answers?

Similar prompts can lead to different answers because AI models respond to wording, context, and settings in different ways. Even small changes can influence the output.

How do you know whether a prompt is working well?

A prompt is working well when the answer is relevant, clear, and close to the intended goal. If the result stays vague or incomplete, the prompt usually needs a clearer purpose or more useful context.

Recommended books and articles on prompting

Prompting helps you guide AI systems more effectively and visibly improve the quality of their output. These books provide a solid foundation for writing, testing, and refining prompts, while the articles demonstrate how prompting has evolved into a serious field within AI, language models, and human-computer interaction. This gives you a clear framework for better understanding prompting and applying it purposefully in content creation, analysis, automation, and knowledge work.

Berryman, J., & Ziegler, A. (2024). Prompt Engineering for LLMs: The Art and Science of Building Large Language Model-Based Applications. Sebastopol, CA: O’Reilly Media. → This book demonstrates how prompting works in real-world applications and helps transform isolated prompt tips into a structured approach.
Chen, B., Zhang, Z., Langrené, N., & Zhu, S. (2025). Unleashing the potential of prompt engineering in large language models: A comprehensive review. AI and Ethics. → This article provides a broad overview of prompting techniques and helps to better understand the strengths, limitations, and areas of application of prompt engineering.
Hernández Gutiérrez, J. A., Conde Díaz, J., Querol, B., Martínez Ruiz, G., & Reviriego Vasallo, P. (2024). ChatGPT: Learning Prompt Engineering with 100+ Examples. Seattle, WA: Amazon Kindle Direct Publishing. → This book makes prompting concrete with many examples and helps you quickly get a feel for structure, precision, and variation in prompts.
Koch, D. (2025). Prompt Engineering in the Enterprise: An Introduction. Wiesbaden, DE: Springer Gabler. → This book places prompting in an organizational context and shows how to use AI instructions to ensure reliability, efficiency, and adoption.
Phoenix, J., & Taylor, M. (2024). Prompt Engineering for Generative AI: Future-Proof Inputs for Reliable AI Outputs. Sebastopol, CA: O’Reilly Media. → This book provides a strong foundation for prompting in generative AI and explains how to move from isolated experiments to consistent output.
Phillips-Wren, G., Laumer, S., & Schryen, G. (2025). Towards using prompt engineering in large language models for decision support. Procedia Computer Science, 266, 1824–1833. → This article demonstrates how prompting is used in decision-making and helps bridge the gap between theory and professional practice.
Sahoo, P., Singh, A. K., Saha, S., Jain, V., Mondal, S., & Chadha, A. (2024). A systematic survey of prompt engineering in large language models: Techniques and applications. arXiv. → This article provides a systematic overview of prompting strategies and demonstrates how widely prompting is now applied across various tasks and domains.
Singh, B. (2025). Building Applications with Large Language Models. Cham, Switzerland: Springer. → This book helps connect prompting to broader AI applications and demonstrates how prompts become part of functioning systems and products.
Vatsal, S., & Dubey, H. (2024). A survey of prompt engineering methods in large language models for different NLP tasks. arXiv. → This article clarifies how prompting differs by task type and helps you better choose which approach suits classification, summarization, reasoning, or generation.

Citation for this article:
Jimmink, J. (2026). Prompting. Retrieved [insert date] from Toolshero.com: https://www.toolshero.com/information-technology/prompting/

Original publication date: April 17, 2026 | Last update: May 4, 2026

Would you like to link to this article? You can!
<a href=”https://www.toolshero.com/information-technology/prompting/”> Toolshero.com: Prompting</a>

Did you find this article interesting?

Your rating is more than welcome or share this article via Social media!

Average rating 4.3 / 5. Vote count: 9

No votes so far! Be the first to rate this post.

Article by:

Job Jimmink

Job Jimmink is Content Manager at Toolshero. He focuses on writing articles and conducting research into management and strategy theories. He also studies at Rotterdam University of Applied Sciences (HES), where he further develops his project management and problem-solving skills. His specific interests lie in procurement management and strategy.