EOLAS - Educational AI Agent

Homepage
Light theme UI
Dark theme UI

Inspiration Our inspiration came from the clear limitations of traditional, one-size-fits-all educational models. We saw a growing gap where students needed personalized guidance but often received static information. We envisioned an AI companion that wouldn't just answer questions but would understand the student's learning pace and style. We wanted to build EOLAS to make adaptive, engaging, and truly personal education accessible to everyone, anytime.

What it does EOLAS is an intelligent, conversational AI learning agent. It acts as a personal tutor that users can interact with via a simple, intuitive chat interface. It goes beyond simple Q&A; EOLAS understands conversational context, allowing students to ask follow-up questions naturally. It can break down complex topics, create study plans, and guide a user's learning journey based on their previous interactions. As an agent, it can reason about a user's needs and provide truly tailored educational support, even explaining its own functions and architecture.

How we built it We built EOLAS on a modern, scalable, and AI-driven architecture:

Frontend: The user-facing application is a cross-platform mobile app built using React Native, ensuring a smooth and responsive experience.

AI Brain: The core intelligence of EOLAS is powered by Amazon Bedrock AgentCore. This allows us to create a true agent with conversational memory, reasoning capabilities, and the ability to execute tasks. We also leverage Amazon Bedrock / Nova for its foundational models.

Backend Logic: We used AWS Lambda for our serverless backend, writing functions in Python. Lambda processes incoming requests from the app, securely interfaces with Amazon Bedrock, and orchestrates the flow of data.

Database & Hosting: We used MySQL hosted on Hostinger to manage user data and learning progress. We also utilized PHP5 and PHPMyAdmin for database administration and handling specific API endpoints.

Security & Version Control: AWS IAM was crucial for managing secure access between our AWS services, and all our code was collaboratively managed on GitHub.

Challenges we ran into Our main challenge was the seamless integration of our diverse tech stack. Configuring AWS IAM roles with the correct permissions for Lambda to securely invoke Bedrock AgentCore was complex and required careful debugging. Another significant hurdle was ensuring the agent maintained conversational context effectively, making the conversation feel natural and adaptive rather than a series of disconnected queries. Finally, bridging our serverless AWS backend with the MySQL database hosted on Hostinger, while also incorporating PHP scripts, required careful API design and network configuration.

Accomplishments that we're proud of We are incredibly proud of successfully building a true AI agent, not just a simple chatbot. Leveraging Amazon Bedrock AgentCore to give EOLAS reasoning and memory is our biggest accomplishment. We are also proud of the clean, intuitive UI we built with React Native, which makes the powerful AI backend accessible to any user. Finally, successfully integrating a serverless AWS architecture with a traditional hosting environment (Hostinger/MySQL) to create a cohesive and functional application was a major win for our team.

What we learned This hackathon was an immense learning experience. We gained deep, hands-on expertise in the Amazon Bedrock suite, especially the power of AgentCore for building stateful, intelligent applications. We also solidified our understanding of serverless architectures using AWS Lambda and Python. More than just the technology, we learned the art of "agent design"—how to structure prompts and logic to move beyond simple Q&A and create an AI that can genuinely assist a user.

What's next for EOLAS - Educational AI Agent This is just the beginning for EOLAS. Our next step is to expand its knowledge base to cover a wider range of academic subjects and curricula. We plan to deepen its personalization by having the agent proactively track a student's progress in our MySQL database and suggest topics based on their learning history. We also aim to equip the EOLAS agent with more tools, such as the ability to access real-time academic databases, create interactive quizzes, and provide even more dynamic learning pathways.