AI-Powered Knowledge Assistant for Drupal

Problem: Drupal users frequently struggle to find specific information across site content, configuration, and documentation because traditional keyword search fails to understand user intent or context. Solution: Develop a decoupled Retrieval-Augmented Generation (RAG) AI assistant. The system will use a Python/FastAPI backend to manage vector embeddings (via databases like Milvus, Pinecone, or ChromaDB) and integrate with a custom Drupal module. This allows users to ask plain-English questions and receive accurate, permission-aware answers grounded in the site's actual data. Deliverables: A Python-based embedding and query API, a custom Drupal module (`ai_knowledge_assistant`) with indexing hooks, an admin configuration interface for managing AI/provider settings, and user-facing chat interfaces (a knowledge block and a dedicated page) that include source citations.