Context Window Consultation — Design Your Context Strategy

I will assess your LLM application and deliver a written context window strategy consultation — covering how to allocate your context window budget across system prompt, conversation history, retrieved context, and user input, how to handle contexts that exceed the limit, and which retrieval or compression strategies are most appropriate for your use case. Context window strategy matters because the model's attention is not uniformly distributed across a full context window — content at the very beginning and end of the window receives stronger attention than content in the middle, making content placement decisions as important as total context volume for tasks requiring the model to use specific information.

The consultation covers your context window budget analysis, allocation recommendation across system prompt, history, retrieved context, and user input, content placement strategy, overflow handling options, and a recommended implementation approach.