ConceptBridge

ConceptBridge – Bridging the gaps in your knowledge with an AI study assistant that helps you master concepts faster, smarter, and with confidence.

Comment

Project Track - 🤖 AI Assistants & Automation

Inspiration

All three of us once struggled to master challenging concepts, and we realized that simply providing answers isn’t enough. Learners need to understand the “why” behind each solution. That insight led us to create ConceptBridge—an AI-driven study assistant that helps students move from confusion to clarity by guiding them through the learning process.

What it does

ConceptBridge processes your own notes or study materials (PDF, Word, or plain text) and automatically breaks them down into focused quiz questions. You then read each question aloud, and our in-browser speech transcription captures your response in real time. Finally, the AI evaluates your answer, highlights any misconceptions, and offers personalized feedback tailored to your needs.

How we built it

We used AWS Amplify to provision and manage both frontend and backend resources with minimal configuration. On the frontend, React combined with Vite gives us lightning-fast development cycles and a flexible TypeScript codebase, all styled with Amplify UI components. File uploads flow into Amazon S3 via Amplify Storage, and our Lambda functions orchestrate AI workflows by invoking AWS Bedrock for question generation. Instead of AWS Transcribe, we opted for the Web Speech API by Google in the browser to achieve low-latency speech-to-text and for ease of integration. To ensure security, each service component operates under its own narrowly scoped IAM role. link We also sit our Lambdas behind API Gateway for the AI so that it only sees authorized requests and credentials never reach the client.

Challenges we ran into

Integrating AWS Bedrock presented an initial hurdle: crafting custom IAM policies and requesting service quotas delayed early progress. We also discovered that AWS Transcribe didn’t meet our responsiveness and integration requirements, which prompted a pivot to the Web Speech API. More broadly, designing a seamless flow—from document ingestion to real-time feedback—required careful planning to keep each component modular and maintainable.

Accomplishments that we’re proud of

We built a fully serverless, end-to-end AI pipeline capable of moving from document upload to personalized feedback in under ten seconds. Our user interface delivers live visualization of both audio input and transcription, allowing learners to see exactly which words have been captured. Additionally, our finely-tuned IAM roles enforce least-privilege access, safeguarding user data and minimizing potential misconfiguration risks.

What we learned

AWS Amplify’s serverless abstractions accelerate development, but a solid understanding of IAM, CloudFormation, and service limits remains essential. We quickly discovered that prompt engineering is an art—small tweaks in phrasing can dramatically improve AI-generated output. Above all, adopting a modular design—splitting tasks into discrete Lambda functions—made testing and debugging far more manageable.

What’s next for ConceptBridge

We plan to expand model options so users can choose between Claude, AWS Kendra, or future AI services. Our roadmap also includes adaptive learning paths that adjust question difficulty based on performance, a React Native client for on-the-go studying, and an analytics dashboard to track progress over time. Finally, we’ll introduce collaboration features, enabling study groups to quiz one another and share insights in real time.

Built With

Share this project:

Updates