FrictionTrace

Architecture Diagram

Inspiration (Why we built it): When a student has a breakdown at school, the traditional system asks: "What is wrong with this child?" It immediately medicalizes the behavior. We were inspired to flip this paradigm completely. We asked: "What is creating friction in the environment?" We built FrictionTrace to look at the environment—crowding, noise, unexpected transitions—instead of putting the blame on the student.

What it does: FrictionTrace is a student-owned platform that logs "moments of friction" throughout the school day. Instead of just creating a list of bad behavior, our autonomous AI Agent acts as an investigator. It analyzes the student's schedule, compares environmental factors (like crowded hallways or loud cafeterias), and connects the dots to find hidden patterns. It then prepares a "Friction Receipt"—a clear, student-owned document that helps adults understand what caused the friction, ensuring the student's voice is heard.

How we built it:

Frontend: Next.js, React, TypeScript, and Framer Motion for buttery-smooth UI transitions. 3D Environment: Three.js and React Three Fiber to build a custom, interactive architectural state machine that visualizes the student's journey and environmental friction. Backend / AI: Python FastAPI and advanced Agentic AI (LLMs). The agent autonomously queries schedule data, compares historical friction moments, and synthesizes environmental patterns without human bias. Challenges we ran into: Translating abstract psychological concepts (like "sensory overload" or "unexpected transitions") into a tangible, visual 3D environment was incredibly challenging. We had to build a custom 3D state machine that dynamically animated a physical architectural model to perfectly match the AI's narrative timeline without dropping frames or breaking immersion.

Accomplishments that we're proud of: We are incredibly proud of "Judge Mode". By pressing 'J' on the keyboard, the user triggers a cinematic 3D storytelling sequence. Instead of just reading text, the user literally watches the AI agent pull back the camera to reveal how a crowded corridor on Monday and a noisy plaza on Tuesday compounded into a breakdown on Thursday.

What's next for FrictionTrace: We want to integrate directly with school SIS (Student Information Systems) so our AI agent can automatically cross-reference friction logs with master schedules, substitute teacher assignments, and even fire drill logs to find even deeper environmental triggers.