Memelang is an AI-optimized query language that significantly reduces token count and model size for LLM RAG. View the code here.