Few-shot prompting
Including one or more worked examples in the prompt so the model copies the pattern instead of guessing it.
Few-shot prompting puts a handful of input-output examples in the prompt before the real request. The model infers the pattern from the examples — the format, the tone, how to handle edge cases — and applies it, which is far more reliable than describing the pattern in prose.
The jump from zero to one example is the largest; further examples only earn their place when each demonstrates a distinct case. Spend them on the boundaries — the ambiguous input, the missing field — not on the easy case the model would have handled anyway.