Fine-tuning uses additional training data to adapt an existing model. Google describes supervised fine-tuning as training an existing model on labelled examples of a task. This is a different operation from supplying instructions and context in a prompt. Source: Google Cloud: Introduction to tuning

Define the behaviour to improve

Choose a specific task, such as classifying incoming enquiries into agreed service categories. Prepare approved examples and describe what a correct result looks like. This gives a proposed training project a testable purpose rather than a general request to make the AI better.

Compare against a simpler starting point

Test the existing model with clear instructions before commissioning training. Keep separate examples for evaluating both approaches and review mistakes as well as successes. This helps establish whether the extra work addresses a demonstrated need in the intended workflow.

Written and updated by Travelgenix. Sources checked on 19 October 2026. Practical advice and examples are Travelgenix editorial guidance.