Inspiration
We were inspired by a simple problem: understanding something is often harder than reading about it. A student can read several pages about photosynthesis, a volcano, the human heart, or molecular structures and still struggle to visualize what is actually happening. We wanted to build something that could turn an explanation into an experience. Orbit was born from the idea that instead of asking AI only for an answer, we could ask it to build a visual explanation.
What it does
Orbit is an AI-powered visualization engine that transforms concepts into interactive 3D learning experiences. A user enters a concept or question, and Orbit analyzes it, breaks it into important components, generates a structured scene description, and uses that description to create an interactive visualization. The experience can combine 3D objects, animations, explanations, narration, and AI-generated quizzes, allowing users to explore a concept instead of simply reading about it.
How we built it
Orbit uses a Flask backend to handle requests and communicate with AI models. We use AI inference to understand the user's concept and convert it into a structured scene representation that the visualization engine can interpret. The frontend uses Three.js to render and animate the generated 3D scenes. This creates a separation between the AI reasoning layer and the visualization layer: the model determines what should be shown, while Orbit's rendering engine determines how it should be shown. The overall pipeline is: User concept → AI analysis → structured scene → 3D rendering → explanation → interactive learning We focused on building the system as a modular pipeline so that new scientific concepts, models, and visualization behaviors can be added without rebuilding the entire application.
Challenges we ran into
One of our biggest challenges was connecting natural-language AI output with deterministic 3D behavior. An AI model can describe a concept extremely well, but a 3D engine needs structured information such as objects, positions, relationships, animations, and interactions. We also had to balance ambition with the 48-hour development window. Instead of trying to build an enormous education platform, we focused on proving the core idea: AI can generate the instructions needed to turn knowledge into an interactive visual experience. Another challenge was making the generated experiences understandable rather than simply visually impressive. The visualization has to support the explanation instead of becoming a distraction.
Accomplishments that we're proud of
We're proud that Orbit goes beyond generating another text response from an AI model. It creates a bridge between AI reasoning and interactive 3D visualization. We built a pipeline where a user's natural-language question can become a structured visual scene, allowing the same system to be applied to completely different concepts. We're especially proud of creating the foundation for an engine that can eventually visualize everything from scientific processes and molecular interactions to engineering systems and mathematical concepts.
What we learned
We learned that building an AI product isn't only about choosing a powerful model. The real challenge is designing the system around the model. Structured outputs, validation, rendering logic, error handling, and user experience are just as important as the AI itself. We also learned that visualization changes the way an AI system can communicate. Instead of only telling someone what happens, Orbit can help them see how and why it happens.
What's next for Orbit
Our next goal is to expand Orbit from a prototype into a general-purpose AI scientific visualization engine. We want Orbit to support more advanced simulations, richer 3D models, voice-driven interaction, uploaded textbooks and PDFs, personalized learning paths, and increasingly complex scientific concepts. Eventually, we want a student to be able to ask something as simple as: “Show me why this happens.” And Orbit won't just answer. It will build the world needed to understand it.
Log in or sign up for Devpost to join the conversation.