Context engineering vs Prompt engineering

Why they differ

Prompt engineering polishes the wording of one request; context engineering manages everything placed in view across many steps, much of it gathered automatically.

Context engineering

Prompting & generation

Choosing all a language model gets to see for a task - instructions, fetched documents, tool results, notes - not just the wording.

Formal

The practice of deciding, for each step of an AI system, what goes into the context window and in what order - system prompt, retrieved documents, tool results, saved notes and earlier messages - within the limit on tokens.

In plain English

Like packing a small suitcase for a trip - what you leave out matters as much as what you put in, and what sits on top gets used first.

In practice

At a Danish software house, a developer sets up a coding agent to load only the task notes and the three most relevant files, and to swap long test output for a short summary before each next step.

Why it matters

In long agent tasks, answers go wrong less from badly worded requests than from missing, stale or cluttered material in view, and untrusted material in view can also steer the agent.

Prompt engineering

Prompting & generation

The craft of wording, ordering and testing the text you send a language model so that it gives useful answers more often.

Formal

The practice of designing a prompt - its instructions, examples, layout and requested answer format - and changing it step by step against test cases, to steer a large language model without changing its weights.

In plain English

Like giving a visitor directions over the phone - you learn which landmarks to mention and in what order, and after each change you check whether the next visitor finds the way.

In practice

A librarian at a public library rewrites the instructions for its book-tip assistant three times, adding two sample replies and “answer in under 80 words”, then reruns fifty saved questions from borrowers to check each version.

Why it matters

It is the cheapest and fastest way to change how an AI feature behaves, but a small change in wording can quietly break cases that used to work, so changes need testing.