Better output from ChatGPT, Gemini, and Claude? Stop with perfect prompts, give better context

For years, "writing a good prompt" was considered the most important skill for anyone wanting to get the most out of AI. That's no longer true. What truly helps language models like ChatGPT, Claude, and Gemini today is context. Here are two techniques that work immediately.

From prompting to context: what's the difference?

A prompt is the instruction or question you give to a language model. Context is the information you provide to help the model understand who you are, what you want to achieve, and for whom the output is intended.

In the past, language models were limited enough that the precise wording of your prompt made the difference between usable and useless output. Models had to be literally directed.

Modern models understand language much better. They require less rigid instructions, but they still can't guess what you mean if you don't tell them. The gap is no longer in understanding the question, but in the lack of background information. Precisely there, context wins over perfect phrasing.

Two techniques to provide more context

Tip 1 Ask the language model to interview you

This may feel unnatural, but it's one of the most powerful techniques. Instead of cramming all the information into one long prompt yourself, you ask the model to ask you the questions it needs to perform the task well.

The model then determines for itself what context it is missing. Your answers precisely fill the gaps that would otherwise lead to vague or generic output. The result aligns much better with what you're looking for, without you even knowing everything you needed to provide.

Example prompt: "Before you start: first ask me the questions you need to perform this as well as possible."

Tip 2 Ask for improvement suggestions after the output

You've had a proposal, email or text written. That's fine, but don't stop there. Then ask the model: "What could you improve about this output?"

Language models are strong in problem-solving. They can elevate something that already exists to a higher level, but only if you give them that opportunity. By asking this question, you force the model to reflect on its own work, and at the same time, you give it more context about what works well and what doesn't in the existing output.

Example prompt: "What could you improve about this answer? Provide concrete improvement suggestions or an improved version."

Why do these techniques work so well?

Both tips have one thing in common: they increase the amount of relevant information the model has when formulating an answer. That's precisely what context does.

With the interview technique, the model actively collects missing information before it begins. With the improvement round question, it reflects on the delivered output with a critical eye and combines that reflection with what it already knows about good text, structure or argumentation.

In both cases, you give the model more input to work with. And more relevant input almost always leads to better, more tailored output.

Summary for immediate application

Context is more important than phrasing: the quality of your output depends on how much relevant information you provide, not on how beautifully your question is constructed.

Let yourself be interviewed: ask the language model what information it needs before it starts. Your answers form the context.

Ask for improvement: once the output is there, ask the model what could be better. You'll get concrete improvements or a stronger version.

Repeat: both techniques are stackable. Use them together for the strongest results.

Frequently asked questions

Why is context more important than a good prompt in AI?

Modern language models have significantly improved in understanding language. As a result, the precise wording of your question matters less and less. What does matter is the underlying information you provide. Without context, the model lacks the information to deliver a tailored answer, no matter how well your prompt is written.

How do I get an AI to interview me?

Send a message like: "Before you begin, first ask me the questions you need to perform this well." The language model will then respond with targeted questions. Your answers will form the context with which it can deliver a much better result.

What is the difference between a prompt and context in AI?

A prompt is the instruction or question you ask. Context is the background information you provide with it: who you are, what your goal is, for whom the output is intended, and what tone is appropriate. Context makes a prompt effective.

Can I ask an AI to improve its own output?

Yes, and it works surprisingly well. After the language model has provided output, you ask: "What could you improve about this?" The model then reflects on its own work and comes up with concrete improvements or an improved version.

Does this work with ChatGPT, Claude and Gemini?

Yes. Both techniques work with all major language models. They are based on how language models generally function, not on specific characteristics of one platform.

Want to get started with AI yourself? Check out our AI Agents e-learning or AI workshops for teams.

• AI-prompts schrijven