Case statistics (total docs, by type/PDF/email, pages, custodians, duplicates, failed-to-process, sizes, dates) should be answered from metadata — not RAG (instant, exact; embeddings are the wrong tool). A first-class Statistics capability; the AI chat should route statistics questions here, not to vector search.
Roadmap stub — implemented in ai_advisor: full design + implementation in Scaia-ai/ai_advisor#58.
Related: #558 (chat routes stats → this), #584 (resources screen).
Case statistics (total docs, by type/PDF/email, pages, custodians, duplicates, failed-to-process, sizes, dates) should be answered from metadata — not RAG (instant, exact; embeddings are the wrong tool). A first-class Statistics capability; the AI chat should route statistics questions here, not to vector search.
Roadmap stub — implemented in ai_advisor: full design + implementation in Scaia-ai/ai_advisor#58.
Related: #558 (chat routes stats → this), #584 (resources screen).