Reference · Glossary
Prompt engineering
Last updated
The practice of **designing instructions, examples, and constraints** so models produce reliable, useful outputs.
When to use
Every serious chatbot, API integration, eval suite, or content workflow — small prompt changes often move quality more than model swaps.
When not to
When missing facts are the bottleneck — better prompts can't replace RAG, tools, or fine-tuning for changing knowledge.
Example
Adding role, audience, format, and "say I don't know if unsure" cuts hallucinations more than switching from mini to full model.