What is prompt engineering?

Prompt engineering is the skill of writing clear instructions for an AI tool so that it gives useful, consistent answers. A good prompt says who is asking, what the task is, what rules to follow, what a good answer looks like and what format to use. The more specific the prompt, the better and more predictable the answer. Most of the skill can be learned in days: be specific, show an example, try again.

Why it matters for a business

Prompting is the cheapest way to improve what a model produces, and the first thing to try before anything more expensive. A short brief often turns an unusable draft into one that needs a single edit. A prompt written down with its own context also travels, so someone else can run it and get the same quality of work. What prompting cannot do is give a model a fact it was never shown. Past that line the fix is putting the document or the record in front of it.

Example

Let’s say a practice manager types “write a job ad” and gets back a dynamic team and a fast-paced environment. A colleague spends 2 minutes adding who the firm is, what the day looks like, the salary band and the opening line of an advert that worked. The second draft goes up with one line cut.

Go deeper

How to write a good AI prompt, and why prompts matter

Related terms

• System prompt
• Few-shot prompting
• Context engineering
• Context window

All terms, A to Z