Solvit!

Solvit! is an AI whiteboard that understands STEM handwriting. It uses a custom logic engine to provide real-time validation and step-by-step guidance, turning a passive board into an active tutor.

Comment

Inspiration

In high school and college, we spent so much time solving problems on digital boards that felt like a waste of space. They let you write, but they don't help you think. Every time we hit a tough equation, we had to stop, switch apps, and manually type it all out or upload a screenshot. That friction is exactly why we started Solvit. We wanted a workspace that feels like a traditional chalkboard but has the brains to solve problems alongside you in real time.

What it does

Solvit! is an intelligent whiteboard that converts your drawings to solutions. Instead of forcing you to type or upload images, Solvit “sees” your drawings. It understands the logic behind every stroke in real-time, turning your sketches into data it analyzes and gives step-by-step guidance.

How we built it

Solvit combines a modern React frontend with an interactive whiteboard (using Konva.js) and a FastAPI backend powered by the Groq AI API. Users can draw mathematical expressions on a canvas, and the system recognizes them, solves them, and explains the solution in real-time.

Frontend React 18 - UI framework TypeScript - Type-safe JavaScript Vite - Fast build tool Konva.js - Interactive canvas rendering KaTeX - Mathematical formula rendering React Konva - React bindings for Konva

Backend FastAPI - Modern Python web framework Python 3.8+ - Programming language Groq API - AI model for recognition and solving SymPy - Symbolic mathematics Pillow - Image processing WebSockets - Real-time communication

Challenges we ran into

  1. Our primary technical hurdle was to find a low-latency model that is capable of recognizing STEM problems accurately. Standard OCR (Optical Character Recognition) struggles with STEM. A handwritten "v" could be the letter v, a velocity vector, or a square root symbol, depending on the physics context. After researching a few models, we narrowed it down to Groq AI API.

  2. During brainstorming, we wanted AI to write on the whiteboard as if it were a real tutor. We couldn't find any existing AI model that can seamlessly overlay on existing whiteboard content. So, we removed the content and modified the UI to show the step-by-step on the side of the whiteboard.

Accomplishments that we're proud of

We are very proud to build a fully functional, intelligent whiteboard as we envisioned in such a short span of time. We built a UI that gave users multiple style options to choose from.

What we learned

  • We learnt how to build a full-fledged application using AI models.
  • HackHayward AI+entrepreneurship track helped us to think about the application from a startup business point of view.

What's next for Solvit!

  1. We would like to execute our marketing plan and make Solvit available for all students.
  2. Personalized AI teaching: Integrate Solvit with the smart assistant to provide suggestions and explanations, giving a personalized learning experience.
  3. Real-Time collaboration: Expand the whiteboard into a shared workspace for multiple users to collaborate in a classroom or study group.
  4. Ecosystem Integration: Integrate with homework systems, learning systems, and learning platform to support a wider audience.

Built With

Share this project:

Updates