Fine-tuning
Specialising a general AI model on your specific data and style.
Fine-tuning takes a pre-trained base model and continues training it on a smaller, task-specific dataset — nudging the model's weights toward your domain, tone, or format.
A customer-service fine-tune might train on thousands of resolved tickets. A legal model might train on firm-specific briefs and precedents. Result: the model responds in the right voice with the right terminology, without lengthy system prompts.
Modern efficient techniques (LoRA, QLoRA) make fine-tuning feasible on consumer GPUs by training only a small fraction of parameters. Fine-tuning isn't always necessary — often prompt engineering or RAG is cheaper and more flexible.
In plain terms
The base model is a brilliant new hire who knows a lot. Fine-tuning is their first 90 days immersed in your company's jargon, processes, and way of working.