Short answer
Prompt engineering is the practice of writing model instructions that produce the required output reliably: stating the role, the task, the constraints, the output format, and one or two worked examples. Adding a filled-in example of the exact structure you expect is often the single change that turns an unreliable step into a dependable one.
Also called: Prompting, Prompt design
Most of the value comes from a short list of habits rather than from secret phrasings. Say what the output must look like and show one. Give the model the information it needs instead of assuming it knows your business. Spell out what to do when the answer is not available, because refusing to guess has to be an explicit instruction. Keep each prompt to one job; a prompt asking for a summary, a classification and a reply at once degrades on all three.
Treated as production code, a prompt needs the same discipline as any other component: kept in version control, changed deliberately, and checked against a fixed set of test inputs before and after each edit. Prompts also drift. A new model version can interpret an old instruction differently, so the test set is what tells you whether an upgrade improved the system or quietly broke one step inside it.
Common questions
Is prompt engineering still a real skill as models improve?
The theatrical part is fading; models no longer need flattery or elaborate role-play to cooperate. What remains is specification. Describing a task precisely, supplying the right context, defining the output format and deciding what should happen in edge cases is ordinary requirements work, and better models raise its value rather than removing it.
Related terms
Large Language Model (LLM)
A large language model is a neural network trained on large volumes of text to predict likely continuations, which lets it write, summarise, classify and translate without being programmed for each task.
Hallucination (AI)
A hallucination is output that is fluent, plausible and false: an invented citation, a policy the company never wrote, a confidently wrong figure.
Guardrails (AI)
Guardrails are the controls that constrain what an AI system may say and do: topic limits, validation of output against an expected structure, blocked actions, spending caps, and filters on both the incoming request and the generated response.
Fine-Tuning
Fine-tuning is further training of an existing language model on a set of example inputs and outputs, so it adopts a specific style, format or classification behaviour by default.
Reading definitions because you are scoping a project? Skip ahead and just ask.