RAG & KNOWLEDGE BASE

Ask a question. Get an answer with sources.

A knowledge base that cites its work. Every answer traced back to the source.

25 articles  |  1,247 embeddings  |  5 categories

Ask a question about Acme Corp, or try one of these:

0 of 25 articles explored
Behind the Build
Stack
RAG PipelineWorkers AIbge-base-en-v1.5Cosine SimilarityLlama 3.1 8BVectorize
Architecture
Query received
Off-topic filter
Keyword fast-match
RAG: embed query (bge-base-en-v1.5)
RAG: vector similarity search
Confidence scoring + reranking
Grounded answer generation
How it works

This is a full Retrieval Augmented Generation (RAG) pipeline. Questions are first checked against pre-built expert answers, then filtered for off-topic content. Relevant queries enter the RAG pipeline: embedded using bge-base-en-v1.5 via Workers AI and compared against 25 pre-embedded knowledge base articles using cosine similarity. The top matches above 60% confidence are retrieved with source citations. Answers are generated by Gemma 4 (with Llama 3.1 fallback) grounded entirely in the retrieved article content. No hallucination: every answer is traceable to a source document.

What you are seeing

A full RAG (Retrieval Augmented Generation) knowledge base for Acme Corp with 25 articles across 5 categories. Pre-built questions return instant expert answers. Free-text questions trigger the real-time RAG pipeline: embedding, vector search, reranking, and grounded generation. Each answer shows inline citations, source documents, and confidence scores based on actual semantic similarity. This is the same architecture used in production enterprise knowledge bases.

Try Chatbots to see this retrieval engine powering a real conversation.

More where that came from.

Back to all demos →