Context Engineering
What is context engineering?
Context engineering is the practice of determining what contextual information an LLM needs to perform a task well and how to provide it effectively. The key principle is giving the LLM exactly the information it needs—nothing more, nothing less.
Context engineering answers questions like: What does the LLM need to know? What files, data, or background information should it have access to? When should that context be retrieved dynamically rather than provided upfront?
This skill is essential alongside prompt engineering when building AI products or using AI as a working tool.
Why can't you just give the LLM everything?
Too much context confuses the model just as much as too little context leaves it unable to complete the task. The quality of the input determines the quality of the output.
Imagine asking an LLM to summarize a meeting and one takeaway is "Send the pipeline report to John by Friday." The LLM might be missing critical context: Which Friday? Which John (if there are multiple Johns)? What is the pipeline report?
You could give the LLM access to your entire employee database, all your reports, and everyone's calendars—but that's like trying to have a conversation in a crowded room. The LLM gets distracted by the noise. Instead, it just needs to know the roles of the people in the room and what the pipeline report is.
How do you provide only relevant context?
This is where techniques like RAG (Retrieval Augmented Generation) come in. Instead of dumping everything into the context, you search for just the relevant information and include that in your prompt.
For example, when using ChatGPT to critique a landing page, you might provide a rich ideal customer profile, course syllabus, student testimonials, and a knowledge base document. Each piece of context helps the LLM understand what to evaluate and why—without overwhelming it with irrelevant information.
Context is king. The more relevant context you can provide (and only relevant context), the better recommendations you'll get.
Learn more:
- AI as a Strategic Thought Partner with UX Implications - All Things Product Podcast with Teresa Torres & Petra Wille
- 21 Ways to Use AI at Work (And Build Your AI Product Toolbox)
- AI Changes Everything (And Nothing At All)
Related terms: