Skip to content
Acme RAG

Corpus

The datasets behind the assistant. Each one carries an access tier that the retriever enforces per role.

This corpus ships with the demo and is already loaded. Live ingestion - fetching a dataset, chunking it, embedding with BGE-M3 and writing to pgvector - runs in the local FastAPI stack, so those controls are disabled here.

Loading datasets...

Retrieval you can put in front of an auditor

A multi-agent assistant over a small company knowledge base. Unlike a general chatbot it enforces who may read what, cites the passage behind every claim, and refuses when the evidence is not there. Three things make that true.