Retrieval-augmented generation, step by step

Bring your own knowledge base. Watch it get retrieved.

Add your own source documents, control how they're chunked, and run a question through the full pipeline: chunking, embedding, cosine-similarity retrieval, prompt augmentation, and generation — with a side-by-side view of a deterministic algorithm versus a real LLM call reading the same context.

Your knowledge base — source documents

Paste in your own material, or keep the sample SOP excerpts below to see the pipeline work out of the box.

Add at least one document to your knowledge base before running the pipeline.

Run the pipeline to see it move stage by stage.
Or click a stage above once it's lit up.