Frizzle uses computer vision and LLMs to accurately grade handwritten math (97% accuracy), flagging uncertain results for human review. Students write on paper; teachers scan work. It delivers rapid, standards-level formative analytics, revealing CCSS mastery. This saves teachers 10-15 hours weekly grading, enables targeted coaching, and reduces math screen time while providing essential classroom data. Trusted by 30+ schools, including Vanderbilt & ASU pilots.