AI Dictionary › Prompting
Context engineering is the discipline of designing and managing the entire body of information a model sees before it answers, not just the prompt. That context includes the system instructions, retrieved documents, tool outputs, prior conversation, memory and examples. The bet is that most model failures are failures of context, giving the model the wrong, missing or badly ordered information, rather than failures of a single instruction.
The practice is about assembling the right context and fitting it into the finite context window. That means selecting what to retrieve, ordering it so the important parts are not lost in the middle, compressing or summarising what is too long, and deciding what to keep in memory across turns. It treats the window as a budget to be spent deliberately.
Context engineering is central to retrieval augmented systems, to agents that accumulate tool results over many steps, and to long assistants that must remember earlier turns. As tasks moved from single questions to multi-step work, managing context became more decisive than wording a single prompt.
The term rose to prominence in 2025, as context windows grew to hundreds of thousands of tokens and practitioners felt that prompt engineering described too narrow a slice of the work. It reframes the craft around the whole context, of which the prompt is only one part.
From our network
AGORÀ Intelligence: Enterprise AI Governance Platform
Govern AI at scale: policies, adoption and measurable results on your data. Built for boards and C-suite.
Visit agora-intelligence.com →From the Agora Intelligence blog
📱 Download the Android app (beta) iOS coming soon
Say what you mean. Get what you need.
Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.