To reduce AI token consumption, it would better to make a memoization tool abou the return value of function calling. And this feature would be optional determined by user, and I will also make it to possible to select which functions to be memoized.