Context engineering
AI Summary
Simon Willison endorses "context engineering" as the term that will replace "prompt engineering" — and argues it has staying power. The problem with "prompt engineering" is that most people's inferred definition is "typing things into a chatbot," a laughably narrow reading that stuck. Context engineering's inferred definition, by contrast, is much closer to the actual complexity: filling the context window with the right information, RAG results, tools, state, history, and compacting strategies.
Willison quotes Shopify CEO Tobi Lutke's definition: "the art of providing all the context for the task to be plausibly solvable by the LLM." Andrej Karpathy amplified the term with a longer thread spelling out the science: task descriptions, few-shot examples, RAG, related multimodal data, tools, state and history, compacting. He calls it "highly non-trivial" and "an art because of the guiding intuition around LLM psychology." Both are influential voices lending credibility to the term's adoption.
The practical implication: practitioners choosing what to call their craft face a real tradeoff between precision and meme-ability. "Prompt engineering" lost the meaning war despite being technically accurate. "Context engineering" frames the discipline as systems design rather than chatbot技巧 — a framing that survives contact with mainstream audiences.
Highlights
- ▸Tobi Lutke (Shopify CEO) coined the definition: context engineering is "the art of providing all the context for the task to be plausibly solvable by the LLM"
- ▸Karpathy on context engineering: science (task descriptions, few-shot examples, RAG, tools, state, compacting) + art (intuition around LLM psychology) = highly non-trivial discipline
- ▸The failure mode of "prompt engineering": the term degraded to mean typing into chatbots; context engineering's inferred definition is closer to the actual complexity
Original excerpt
Sponsored by: Honeycomb — AI agents behave unpredictably. Get the context you need to debug what actually happened. Read the blog
The term context engineering has recently started to gain traction as a better alternative to prompt engineering. I like it. I think this one may have sticking power.
I really like the term “context engineering” over prompt engineering. It describes the core skill better: the art of providing all the context for the task to be plausibly solvable by the LLM.
+1 for "context engineering" over "prompt engineering". People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context…
20 more articles in this vault.
Import the full Context Engineering vault to Burn 451 and build your own knowledge base.
Content attributed to the original author (Simon Willison). Burn 451 curates publicly available writing as a reading index. For removal requests, contact @hawking520.