A person asks a large language model to develop a product description for a laptop.

The scenario describesLeast to mostprompting. This technique involves breaking down a complex task into smaller, manageable sub-problems and solving them sequentially. In this case, the user starts with a basic request and progressively adds layers of complexity—details, context, and restrictions—to guide the AI toward a sophisticated final output. It is essentially a strategy of "building up" the prompt complexity until the model has enough specific information to meet the high-level requirement.

Unlike "Chain of Thought" (COT), which focuses on the AI showing its internal reasoning steps for a single logic problem, "Least to most" is about the user-led structural decomposition of a task. It is highly effective for creative or technical writing where a "zero-shot" (single try) approach often yields generic results. By refining the prompt iteratively, the user ensures the AI understands each constraint before moving to the next level of detail. In practical applications, this technique is used to "warm up" the model’s context window with specific domain data, ensuring that by the time the final description is generated, the AI is fully aligned with the technical specs and brand voice required for the laptop.