Few-Shot Prompting
Including several examples in the prompt so the model infers the pattern you want.
Few-shot prompting teaches by demonstration inside the context window, with no training involved. It is the highest-leverage prompting technique for consistency, especially on classification and formatting. Choose examples that cover the edge cases, not just the easy ones — the model will copy whatever you show it, including the gaps.
In practice: Five labelled tickets in the prompt, and category accuracy jumps without any fine-tuning.