Few-shot Prompting
In one line
Few-shot prompting includes a small number of example inputs and outputs in the prompt itself, letting the LLM imitate the desired format or style without any retraining.
Going deeper
Few-shot prompting embeds 2–5 input/output examples directly in the prompt so the model mimics the pattern. Nothing is being retrained — the model just imitates what it sees in context. This behaviour is sometimes called in-context learning.
For marketers it is the cheapest way to enforce brand voice quickly. Three or four well-chosen examples can transform output style, often making fine-tuning unnecessary.
The downsides are real, too. Long example sets eat tokens and chew up the context window. Knowing when few-shot is enough versus when you genuinely need fine-tuning is a practical judgement call.
Related terms
Prompt Engineering
Prompt engineering is the practice of crafting inputs that steer an LLM toward better outputs — a way to dramatically change result quality without retraining the model.
LLMSystem Prompt
A system prompt is the instruction sent to an LLM before any user message, defining the assistant's role, tone and rules — effectively the AI product's character.
LLMChain-of-Thought
Chain-of-Thought (CoT) prompting asks the LLM to walk through intermediate reasoning steps before giving a final answer — a simple change that meaningfully improves accuracy on harder problems.
LLMFine-tuning
Fine-tuning takes an already pretrained LLM and trains it further on a narrower dataset to specialise it for a domain, task or voice — the most common path for adapting an LLM to your own data.
LLMContext Window
The context window is the maximum number of tokens an LLM can take in at once — it defines how much content the model can consider in a single prompt.