Question 12
Domain 1: Agent Architecture, Design, and DevelopmentWhat architectural approach would best manage context for these large-scale cases?
Correct answer: B
Explanation
A hybrid context architecture fits large-scale cases because it combines targeted retrieval with short-term memory and a compact overview. RAG “fetches relevant historical documents based on the query,” while a conversation buffer preserves recent turns and a case summary maintains overall context, all within a standard context window.
Why each option is right or wrong
A. Use the largest available context window model (200K+ tokens) and fit as much case history as possible, summarizing older documents to save space.
B. Implement a hybrid context architecture with a retrieval system (RAG) that fetches relevant historical documents based on the query, combined with a conversation buffer for recent context, and a case summary for overall context, all fed into a standard context window.
Large-scale matters here because a single prompt window cannot reliably hold the full case history, so the architecture has to separate long-term storage from short-term working memory. A hybrid design is the only option that simultaneously supports targeted recall of prior documents, retention of the most recent exchanges, and a compact global summary, all of which are then injected into the model’s finite context window for generation.
C. Fine-tune a model on each case's documents so the model internalizes all case knowledge and doesn't need external context.
D. Split the case into chronological chunks, process each chunk separately, and combine the results from all chunks.