Context engineering

The practice of designing which information reaches an AI model, in what order, and in what format, so the model can do useful work.

Writing a good prompt is one small piece of the puzzle. Context engineering covers the full system: which documents to retrieve, how to compress them, what conversation state to carry forward, which tools to make available, what to store in memory, and how to stay within the model's token budget. Think of it like preparing a briefing packet for a new consultant. The quality of their advice depends on what you include, how you organize it, and what you leave out.

Builder example

When an AI product gives a bad answer, the cause is usually wrong information reaching the model, not wrong wording in the prompt. A customer-support bot that retrieves last year's return policy instead of this year's will confidently cite the wrong rules. Context engineering controls that information pipeline so the right material arrives, structured clearly, every time.

You ask an AI to draft a client email. It produces three paragraphs of corporate filler because it has no examples of how you write to clients.

Give it two or three of your previous emails as examples. The difference in output quality is immediate.

Common confusion: Adding more text to the input often makes a model perform worse, especially when the text is unstructured or irrelevant. Context engineering means selecting and organizing information, not piling it on.