Learn AI · Level 3 · Tech and tools · step 3 of 16 · word 3 of 3
What is context engineering?
Context engineering is choosing everything an AI model sees before it answers, such as instructions, documents, past messages and tool results, so it has exactly what it needs.
In plain English
A model only knows what is in front of it at that moment: its context window. Prompt engineering is about wording one request well. Context engineering is the bigger job of deciding what else goes in: which rules, which documents, which parts of the conversation, which tool results.
The term caught on in 2025, as AI agents started running long tasks with many steps.
An example
A customer-support bot gets the company’s refund policy, the customer’s last three orders and a summary of the chat so far. It does not get the whole policy manual or every past conversation, which would bury the important parts.
Watch out
More isn’t better. Stuffing the context with everything makes answers slower, more expensive and often worse, because the key detail gets lost.
Learn AI path · Level 3 · Tech and tools · step 3 of 16
Done? Next stepHow to tell if an AI system actually works: evals in plain English →See all of level 3