GlossaryStage 1 of 4 · Prompt
Prompt engineering
Prompt engineering is the practice of writing instructions for a language model so that it returns a useful answer: what to ask, what examples to show, what format to return. In a business it is the first stage of working with AI, where each person writes their requests from scratch.
How prompt engineering works
A prompt is everything the model reads before it answers. Prompt engineering shapes it: a clear task, the role the model should take, a few examples of a good result and the format of the answer. For example, asking for "three subject lines under 50 characters for a webinar invite" works better than "write subject lines".
Prompt engineering vs context engineering
Prompt engineering improves a single request. Context engineering decides what the model knows before any request: the company's decisions, documents and history. A well-written prompt still fails when the model knows nothing about your business, which is why teams that stop at prompts get uneven results that depend on who is asking.
Is prompt engineering a real job?
It can be, and some companies hire for it. In most small teams, though, it is a skill everyone who works with AI needs, not a separate role. The bigger gain usually comes from the next stages: shared knowledge, checked results and work that runs on its own.