Prompt engineering

Prompt engineering is the practice of writing clear, structured instructions to an LLM so it produces better output. It is not a separate job for most people—it is the skill of describing what you want in a way the AI can execute.

In one sentence

Prompt engineering is just writing good instructions: the clearer you are, the better the result.

What it's for

Everyone who uses an LLM does prompt engineering, whether they call it that or not. The difference between a vague request (“write me an email”) and a specific one (“write a polite email to my dentist rescheduling from Tuesday to Thursday, under 100 words”) is the difference between garbage and gold.

How to think about it

Think of it like delegating to a new coworker. You wouldn’t say “fix the report.” You’d say “Review section 3 for typos, update the sales chart with Q3 numbers, and email the final version to the team by 5pm.” Same principle.

Common misunderstandings

You do not need courses, certificates, or jargon. Most effective prompts use three things: role (“act as a...”), context (“for a manager who...”), and format (“in 3 bullet points”). That’s it.

If you want to try it

Take an email you wrote last week. Ask ChatGPT to rewrite it twice: once with a vague prompt, once with a specific role + context + format prompt. Compare the two. The difference is prompt engineering.

Sources

• Learn Prompting (open-source guide)
• Anthropic: Prompt Engineering overview