Fine-tuning
Fine-tuning is the process of further training an existing AI model on a set of your own examples so that it adopts a particular style, format or specialised behaviour more reliably.
Why it matters for a business
Fine-tuning changes the model itself, while prompting and retrieval change what the model is given. It can help when you need a very consistent output style, a narrow classification task done at high volume, or a tone that is hard to describe in instructions. It also takes more effort: you need many good examples, a way to measure results and a plan to repeat the work when things change.
An example
A brokerage in Dubai wants every listing description written in its house style. It first tries a detailed prompt with a few sample descriptions, which gets most of the way. Only if the team later needs thousands of descriptions with very little editing would it consider fine-tuning a model on its approved past listings.
Common mistake
Businesses reach for fine-tuning to teach the model facts such as prices or stock. Facts change, and a fine-tuned model does not update itself. Use retrieval-augmented generation for knowledge and good prompt engineering for behaviour first. Consider fine-tuning only when those two have been tested and still fall short.
The daily skill, in your inbox
One email each morning with the new drop. Unsubscribe in one click.
