Agent designFree→ Claude / ChatGPT
RAG System Design Review
Review a retrieval-augmented design for quality, security and cost.
What you get: A review covering chunking, retrieval, grounding, evals and tenant isolation.
Review my RAG design. Use case: {{use_case}}. Corpus: {{corpus}} (size, types, update rate). Users and tenancy: {{tenancy}}. Current design: {{design}}.
Assess and recommend on: chunking and metadata; embedding model choice and re-embedding plan; hybrid retrieval and reranking; filters for tenant isolation (security); grounding prompt and citation format; handling no-answer cases; freshness and deletion; evaluation (golden set, recall@k, faithfulness); latency and cost; and failure modes. Give a prioritized list of 10 improvements with the metric each should move.