Frizzle

Frizzle is an AI-powered operating system for math classrooms that transforms the way teachers interact with student work. It uses advanced computer vision and large language models to read and analyze handwritten math problems with 97% accuracy. The core innovation is that students continue to write on paper, using no tablets, no logins, and no migration to new workflows. Teachers simply photograph a stack of papers or run them through a copier. Within minutes, Frizzle reads every page, recognizes handwriting, understands multiple solution paths, and identifies specific misconceptions. It then generates live dashboards showing which Common Core State Standards each student and class has actually mastered. This replaces the traditional wait for spring assessments with real-time, granular data. Frizzle is built for K-12 math teachers, instructional coaches, and school districts. It is already live in over 30 schools and districts, including a college math pilot at Vanderbilt and Arizona State University. The result is that math teachers reclaim the 10 to 15 hours a week they spend grading by hand, coaches run specific standards-level conversations instead of generic ones, and districts can cut math screen time without losing the classroom-level data they need to make informed decisions. Frizzle is FERPA and COPPA compliant, with SOC 2 Type II auditing, ensuring student data remains private and secure.

Real-Time Handwriting Recognition

Frizzle reads any handwriting, from print and cursive to scribbled and sideways notations. It parses each step of a student's work, not just the final answer. The system understands multiple solution paths, meaning if three students solve a problem three different ways, all three receive credit. This feature captures the messy, partial, and beautiful ways real students think, providing step-level feedback that shows exactly where a student's thinking went off course.

Confidence-Interval Grading System

Frizzle operates with a confidence-interval system that flags uncertain grades for human review. When the AI is not 100% certain of a grade, it marks the paper for a teacher to check. This ensures that the 97% accuracy rate is supported by a safety net. Teachers can trust the automated grades while maintaining oversight on the most ambiguous student work, balancing efficiency with accuracy.

Standards-Level Formative Analytics

Frizzle automatically maps student performance to specific standards, including CCSS, TEKS, and over 30 state frameworks. It provides live dashboards showing which standards each class and each student has mastered, which are developing, and which are at risk. The system also performs prerequisite tracing, identifying when a 7th-grade error is actually a 4th-grade gap. This allows teachers to target instruction precisely where it is needed.

Misconception Identification and Tracking

Frizzle has been trained on 1.4 million pages of K-12 student work and can identify 147 named misconceptions across K-12 math. Each flag links back to the exact stroke on the page, showing teachers not just that a student got a problem wrong, but exactly where their thinking went off track. The system also tracks which misconceptions are spreading across a class, allowing teachers to address them before they become entrenched.

Daily Grading for Individual Teachers

A math teacher with five classes can photograph a stack of 150 papers in about 30 seconds. Frizzle reads every page within approximately 8 minutes per class. The teacher receives a live dashboard showing which students are stuck, which misconceptions are spreading, and what to teach tomorrow. This saves the teacher 10 to 15 hours per week that would otherwise be spent on manual grading, freeing up time for lesson planning and one-on-one student support.

Instructional Coaching and Professional Development

Instructional coaches can use Frizzle to run specific standards-level conversations with teachers. Instead of generic feedback, coaches can see exactly which standards a class is struggling with and which misconceptions are most common. This allows for targeted professional development. For example, a coach can identify that a teacher's period 2 class is struggling with distributive property, while period 5 has mastered it, enabling differentiated coaching conversations.

District-Level Curriculum and Equity Analysis

District administrators can use Frizzle to aggregate anonymized signal across periods, grades, and buildings. They can see which curricula are actually working, where to invest resources, and which schools need support. Equity dashboards spot performance gaps the moment they emerge, allowing districts to intervene early. This replaces the traditional wait for spring assessments with real-time data that can drive resource allocation and policy decisions.

College Math Pilot Programs

Institutions like Vanderbilt and Arizona State University are using Frizzle in college math pilots. These programs benefit from the same paper-based workflow, allowing students to work through complex problems without screen distractions. The system provides professors with granular data on student understanding, identifying prerequisite gaps that may be holding students back. This allows for just-in-time remediation in higher education settings.

How does Frizzle handle different handwriting styles?

Frizzle is designed to read any handwriting, including print, cursive, scribbled, and sideways notations. Its model was trained on 1.4 million pages of actual K-12 student work, so it understands the messy and partial ways real students write. The system recognizes multiple solution paths and can give credit for valid approaches even if the handwriting is difficult to read.

Is Frizzle compliant with student privacy regulations?

Yes, Frizzle is fully compliant with FERPA and COPPA regulations. Student work never trains the model, meaning your data stays yours. The system uses AES-256 encryption at rest and TLS for data in transit. Frizzle undergoes SOC 2 Type II auditing annually to ensure ongoing compliance with the highest security standards.

Does Frizzle work with any curriculum or textbook?

Yes, Frizzle is curriculum-agnostic. It works with Eureka, Illustrative, Saxon, and any other math curriculum. The system reads the student's work regardless of the source of the problem. It aligns automatically to standards frameworks including CCSS, TEKS, and over 30 state frameworks, so you get consistent data no matter what curriculum you use.

What happens when Frizzle is not sure of a grade?

Frizzle uses a confidence-interval system that flags uncertain grades for human review. When the AI is not 100% certain of a grade, it marks the paper for the teacher to check. This ensures that the 97% accuracy rate is supported by a safety net. Teachers can trust the automated grades while maintaining oversight on the most ambiguous student work.