Frizzle, founded in 2025 by Abhay Gupta and Shyam Sai and backed by Y Combinator, lets a teacher photograph a stack of handwritten math papers and reads every page, identifying which students are stuck, what misconceptions are spreading, and what instruction the class needs next, work that otherwise consumes hours of manual grading a week.
Its model reads print, cursive, and messy handwriting at 97% accuracy, trained on 1.4 million pages of real student work, and analyzes step-level reasoning and partial credit rather than only the final answer, tagging against 147 named mathematical misconceptions and tracing prerequisite gaps back to their root. The platform works with Eureka, Illustrative Math, Saxon, and more than 30 state frameworks rather than locking a school into one curriculum.
Frizzle is free forever for individual teachers, with school and district plans adding admin dashboards, single sign-on, and standards reporting. For a math teacher who wants handwritten student work graded and analyzed for the reasoning behind it, rather than scored alone, Frizzle's step-level, misconception-aware grading addresses that directly.
