[mikert89]

chunk the documents, use contextual embeddings, put into the vectordb in postgres