HAWKI RAG is the customised retrieval deployment used in the HAWKI project. It keeps the Laravel application and FastAPI bridge you already know, but rebrands the end-user experience and Docker stack, highlighting the combo of Qdrant + Neo4j + the HAWKI RAG pipeline.
HAWKI RAG is designed for fast retrieval over crawled HAWKI content. By default it uses
bge-m3 for embeddings and llama3:8b / llama3.1:8b for grounded answers.
