← Back to Demos
RAG Mini
Simple Retrieval-Augmented Generation pipeline. Ask questions about sample documents and see how chunks are retrieved and answers are generated.
Simulated demo. Chunk retrieval uses keyword matching to approximate vector similarity search. Answers are pre-written for common questions. The real IDP System uses pgvector embeddings for similarity search and live LLM generation for freeform answers.
Document Corpus
Query→Embed→Search→Retrieve→Generate
Ask a question about the loaded documents.