Prompt engineering is the practice of designing a model's input: the instructions, the context, and the output format. The aim is a reliable result, not one lucky answer.
It still matters because capability is not the same as intent. A strong model writes fluent text with or without a good prompt. It cannot guess your audience, your edge cases, or what to do when the context lacks the answer.
The prompt is the cheapest thing you can change. Editing it takes seconds, with no training run and no new data.
Wording changes cost and consistency. A production prompt runs thousands of times, so the effect of a small edit multiplies.
Better models still write confident wrong answers. A clear prompt makes good output more consistent and wrong answers less likely.
Rewriting in plainer words…
This answer doesn't lend itself to a diagram - it reads best . No credits were charged.
When do you stop tuning the prompt and move to retrieval or fine-tuning?Prompting fixes intent and format. Missing knowledge points to retrieval. Behavior the prompt can't hold reliably or cheaply points to fine-tuning.
Prompt engineering is designing the model's input, meaning the instructions, the context and the output format, so you get a reliable result and not one lucky answer.
It still matters because capability is not the same as intent, and a strong model writes fluent text whether your prompt is good or not.
What the model can't do is guess your audience, your edge cases, or what you want when the context doesn't have the answer.
The prompt is also the cheapest thing to change, because an edit takes seconds and needs no training run and no new data.
In production a prompt runs thousands of times, so a small wording change adds up in both cost and consistency.
Better models still write confident wrong answers, so a clear prompt makes good output more consistent and wrong answers less likely, but it won't remove them.
Weak answers to avoid
Reduces it to magic phrases and tricksThat sounds dated. Define it as specifying instructions, context and output format so the result holds up across many inputs.
Claims better models make it obsoleteCapability doesn't tell the model your audience, edge cases or fallback. Say what still has to be spelled out, and that the prompt is the cheapest thing to change.
Judges a prompt by one good outputThe goal is reliability across thousands of runs. Say you test on many real inputs, because one lucky answer tells you little.
Presents a good prompt as a hallucination fixA clear prompt makes wrong answers less likely, not impossible. Say that plainly instead of promising the prompt alone makes output correct.
Interview lens
Likely follow-ups, what you can say, and the weak answers to avoid.
