Fine-tuning is the process of taking an already-trained AI model and adjusting it with your own set of examples, so it specializes in a specific task, format or style. Instead of training a model from scratch (extremely expensive), you start from a ready model and "tune" it with your data.
Fine-tuning vs RAG
This is the most common confusion. RAG gives the model access to external knowledge at answer time (good for facts that change). Fine-tuning changes the model's behavior (good for style, format and repetitive tasks). In most enterprise cases, RAG solves it first; fine-tuning comes in when the response pattern must be highly consistent.
When it is worth it
Fine-tuning makes sense when you have many quality examples of the desired task and need consistency or lower cost per answer. It does not make sense for "teaching facts" to the model, that is RAG's job. The rule of thumb: start simple (prompt + RAG) and only invest in fine-tuning when the problem demands it.