Ridiculous. Are you going to evaluate the performance of an AI agent by how many tool call it made?