Applied AI·Fine-tuning and customisation
you were about to fine-tune and found the real fix was a clearer prompt, retrieval, and the eval set you did not have.
The decision rule
Draft summary, pending review
In order: prompt engineering first, then retrieval for knowledge, then fine-tuning only when both plateau and you hold hundreds to thousands of good examples. Most teams who think they need fine-tuning need a better prompt and an eval set.