Direct answer
This is the answer library for SavantCat. It collects short, direct answers to the questions small and medium businesses actually ask about building an enterprise knowledge base and running AI customer service in China — each one stating what it is based on, so a reader (or an AI model) can check the claim rather than take it on faith.
How this library is organised
Answers are grouped into clusters. Each cluster has a hub page that explains the whole subject, and a set of atomic answers that each take one question and answer it completely. Every page lists its sources, and every claim about a regulation refers to the clause it comes from.
Clusters
AI Customer Service & GB/T 47746—2026: Six Answers
How China's national standard for human–AI customer service collaboration works, what it requires, and how a small business gets through it.
What is GB/T 47746—2026 and who does it cover? · GB/T 47746: does my company have to comply? · Where does the figure of 53 mandatory items come from? …
Enterprise Knowledge Base: Six Decisions That Decide It
How to build an enterprise knowledge base that AI can actually answer from — chunking, retrieval, evaluation, and publishing an answer layer.
Why do enterprise knowledge base projects fail? · How should you chunk documents for a knowledge base? · RAG vs fine-tuning for an enterprise knowledge base …