RAGDC isn't just a ChatGPT wrapper — we ship a full enterprise-grade RAG stack. This guide walks through the 4 core modules: Knowledge Base, Retrieval, Generation, Integration — so you can judge which scenarios fit.
Key takeaways
A knowledge base is an isolated document container. Split by business line (contracts, policies, product manuals), by client (one per customer), or by department. Supports PDF, Word, PPT, Excel, Markdown — or connect to SharePoint sites with files staying in your Microsoft cloud. Each workspace allows multiple KBs; the count limit depends on your plan tier.
When a question comes in, the system doesn't feed the entire document to the AI (slow and expensive) — instead it uses hybrid retrieval to find the most relevant chunks, then a dedicated reranker picks the closest matches. This step directly determines answer quality — our approach is 30-50% more accurate than naive embedding search.
With retrieved chunks in hand, the AI model (Claude / GPT / Gemini selectable) generates a natural-language answer grounded in those paragraphs. The system instructs the AI to "only answer from provided context, say I don't know otherwise" — the key difference from bare ChatGPT. Answers auto-cite sources, clickable to view originals.
Via our API, engineers can embed AI Q&A anywhere: website chat widget, internal portal, Slack / Lark bot, mobile app. The API is pay-per-call, with all models managed by us — no infrastructure deployment on your side. Detailed info on the API Reference page.