Many companies want to experiment with AI, but they also need to keep control over their data, infrastructure and model choices. This short talk presents a minimal Retrieval-Augmented Generation prototype built with open data, an open-source database and an open model.
The session walks through the essential RAG pipeline: loading data, creating chunks, generating embedding, storing them in a searchable database, retrieving relevant context and producing grounded answers. The goal is to show how companies can start exploring private AI with a small, understandable and reproducible open-source setup.
