ModelRefs / Prompt Engineering — AI Glossary
Prompt Engineering — AI Glossary
The discipline of designing LLM inputs to reliably produce desired outputs. It is the cheapest lever before retrieval, fine-tuning, or model upgrades.
Overview
Effective prompts combine clear role framing, structured output schemas, few-shot examples, explicit constraints, and chain-of-thought elicitation. It is the cheapest lever before retrieval, fine-tuning, or model upgrades. System prompt placement matters — models weight it above user messages.
Reference details
| Topic | prompting |
|---|---|
| Last reviewed | 2026-06-24 |
Related terms
Example: The same request, made checkable
“Summarise this support thread” returns a different shape every run, so nothing downstream can consume it. “Return JSON with keys issue, root_cause, resolved (boolean), next_action. If the thread does not state a root cause, set root_cause to null — do not infer one.” now has an explicit failure path, which is what makes the output testable rather than merely nicer.
Commonly confused with
Prompt engineering is not phrasing tricks. Most of the durable gain comes from structure — a clear role, an output schema, representative examples, and stated behaviour for the cases the model would otherwise guess at. Prompts that depend on a magic incantation tend not to survive a model upgrade.
When to use it
Reach for it when:
- Before any heavier intervention: it is the cheapest and fastest lever
- The model can already do the task but is inconsistent about form
- You need behaviour you can change without redeploying a model
Reach for something else when:
- The model genuinely lacks the knowledge — that is retrieval or fine-tuning
- You are pasting an ever-growing rulebook into every call; that cost recurs forever
- The requirement is a hard guarantee, which a prompt cannot provide
Referenced by
This term is used by the following ModelRefs references:
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Prompt Engineering — AI Glossary.
Frequently asked questions
What is Prompt Engineering?
The discipline of designing LLM inputs to reliably produce desired outputs.
What concepts are related to Prompt Engineering?
Closely related concepts include chain of thought, function calling, few shot, system prompt.