Glossary / AI & data

Fine-tuning

Fine-tuning takes a pre-trained model and continues its training on your own examples, teaching it your domain's language, format, or judgment. It shines for repetitive, well-defined tasks with plenty of examples — classifying your tickets, matching your writing style. For most business needs, though, good prompting plus RAG gets there cheaper and stays easier to update; fine-tune when evidence says you must.