# Chain-of-Thought

> Chain-of-thought is a prompting technique that asks a model to reason step by step before giving a final answer, which improves accuracy on multi-step problems like maths, logic and planning.

*Source: https://www.lazlosoftwaresolution.com/glossary/chain-of-thought*

By making intermediate reasoning explicit, chain-of-thought reduces errors on tasks that need several steps. In production it is used carefully — the reasoning can be hidden from end users while still improving the quality and reliability of the final output.