efficiency

RecursiveMAS reduces token usage by 75% and speeds up multi-agent AI inference, making it a scalable solution.