Fine-tuning
Continuing to train a base model on your own examples so it adopts a style or skill.
Fine-tuning is most useful when prompt engineering and RAG aren't enough — for niche tone, structured outputs, or compressing many examples into the model itself.