Document Management Assistant

An internal chatbot for authorized users to query case files from the GDE (Electronic Document Management) system. The assistant understands the status of each case, summarizes its content, and answers specific questions (who signed it, what documents it contains, what step it's on) while always citing its source. It uses RAG (Retrieval-Augmented Generation): each case's documents are indexed and stored as embeddings in PostgreSQL with pgvector, and on every query the system retrieves only the relevant fragments from those documents to feed the model as context, instead of letting it answer from memory. This keeps every answer grounded in the case's real documentation rather than invented information.