Production Grade Deterministic RAG pipeline - Inquire AI Demo

InquireAI is an enterprise RAG knowledge assistant for asking grounded questions over internal documents. It combines structure-aware document ingestion, hybrid retrieval, permission-aware context, and citation-backed answers in one workflow. Features: 1. Upload PDF, DOCX, XLSX, Markdown. 2. Structure-aware parsing and chunking via Docling 3. Hybrid retrieval: sparse BM25 + dense vector search with reciprocal-rank fusion 4. Cross Encoder re-ranking 5. Contextual chunk enrichment (Groq-generated summaries at ingestion time) 6. Contextual query rewriting using session history 7. Multi-hop retrieval with sufficiency checks 8. Grounded answers with structured citations and citation validation 9. Graceful abstention when evidence confidence falls below threshold 10. RBAC filtering enforced at retrieval time, not after generation 11. Guardrails: Prompt Guard injection detection, 12. Presidio PII screening, Llama Guard content safety 13. FastAPI backend with Logfire observability and LangSmith agent tracing 14. React frontend with user impersonation, conversation management, and document ingestion panel Link: #rag #production #ai #langchain #agenticai #python #hybridsearch #database #backend