Knowledge Assistant

This is a RAG chatbot (Retrieval-Augmented Generation): it only answers from real technical documentation — it doesn't make things up. It finds the most relevant passages in the docs — the "retrieval" step — and uses them to generate an answer. If the answer isn't in the docs, it says so instead of guessing.

• ChunkingDocs split by heading section, not fixed length — each chunk stays semantically coherent
• Hybrid RetrievalVector similarity search + full-text search, fused via Reciprocal Rank Fusion — not similarity alone
• Strict GroundingThe model answers only from retrieved context, never from its own knowledge — no exceptions
• Rate LimitingPublic endpoint protected per-IP against cost abuse
• Input Validation (Zod)Every request validated at runtime before hitting retrieval/generation — no untyped input passes through