Doc Chatbot: RAG vs fine-tune vs hybrid – decision framework

Review-style guidance: - Pure RAG: Fast to build, transparent, easy updates; relies on retrieval quality. - Fine-tune: Strong for stylized or domain-specific generation; harder updates. - Hybrid: Light fine-tune for style/refusals + RAG for facts. Questions: What tipped you to fine-tuning (e.g., stable doc set, brand voice)? Any benchmarks you used to justify the cost? Include your maintenance costs over 6-12 months.