Inspiration
Learning is not just about memorizing information; it is about understanding concepts and how they connect. Many students struggle to identify gaps in their understanding and know what to practice next. We wanted to build a learning platform that goes beyond static explanations and helps learners actively develop conceptual understanding through personalized practice and feedback.
What it does
ConceptBridge transforms learning into an interactive experience. Learners can enter a topic, select a difficulty level, and generate an AI-powered lesson with explanations, concept connections, and practice challenges. The platform evaluates submitted answers, identifies potential knowledge gaps, provides feedback, and generates follow-up challenges. It also offers visual concept networks, learning activity tracking, and a spaced-review list to support revision.
How we built it
We built ConceptBridge using Python and Streamlit for the application logic and user interface. Google's Gemini API, accessed through the Google Gen AI SDK, powers lesson generation and answer evaluation. Pydantic structures the AI responses, while SQLite stores lessons and practice attempts locally. Mermaid.js renders concept diagrams and knowledge networks. The application is deployed using Streamlit Community Cloud, with the source code maintained on GitHub.
Challenges we ran into
One of our key challenges was handling the variability of AI-generated responses and making them suitable for a structured learning experience. We also encountered API availability, quota, and temporary service errors during development. Another challenge was bringing multiple features—lesson generation, answer evaluation, visual diagrams, and progress tracking—together into a single application while keeping the experience understandable for learners.
Accomplishments that we're proud of
We built and deployed a working learning platform that combines AI-generated lessons, concept visualization, answer evaluation, adaptive follow-up challenges, and learning progress features. We are particularly proud of connecting explanation with practice and feedback, helping learners do more than simply read information. We also made the project publicly accessible and documented its source code on GitHub.
What we learned
We learned how to integrate an AI API into a web application, structure model responses with Pydantic, store application data using SQLite, and build interactive interfaces with Streamlit. We also learned the importance of handling external service failures, protecting API credentials, testing features in a deployed environment, and designing AI experiences around a real user problem rather than AI generation alone.
What's next for ConceptBridge
Our next goal is to make learning more personalized over time. Potential improvements include learner accounts, cloud-based progress synchronization, more detailed learning analytics, improved personalization based on past performance, and expanded review strategies. We also want to evaluate how effectively ConceptBridge helps learners understand and retain concepts through real-world user testing.
Log in or sign up for Devpost to join the conversation.