We have introduced a robust system for tracking AI usage across all Gemini calls, allowing for granular monitoring per school, user, model, and feature. This update includes configurable AI policies that support warnings or usage blocks based on daily, weekly, or monthly thresholds. Additionally, we implemented several cost-saving optimizations, such as request caching and utilizing Gemini 'Flex' tier for non-urgent background tasks, to ensure more efficient AI operations without sacrificing quality.