Fine-Tuning.

Fine-tuning takes a pre-trained model—which has already learned general language understanding from massive datasets—and continues training it on a smaller, task-specific dataset. This updates the model's weights to specialise its capabilities for the target domain while retaining the general knowledge acquired during pre-training.

Fine-Tuning vs. Prompt Engineering

Prompt engineering instructs the model what to do at inference time without changing its weights. Fine-tuning actually changes the model's parameters to make certain behaviours more natural and consistent. Fine-tuning is more expensive but produces more reliable specialisation; prompt engineering is more flexible and iterative.

Applications in Marketing

AI writing tools, SEO scoring models, and ad copy generators are often fine-tuned versions of foundation models. A model fine-tuned on high-quality SEO content learns to write in a style that tends to perform well, while a model fine-tuned on ad copy data learns conversion-focused language patterns.