Polarx_PixelPirates_AntonTrack

The logo

Inspiration

Comparing policies whether it be insurance, financial plans, or service agreements can be confusing and time-consuming. Many policies are written in dense, technical language, making it difficult for users to clearly understand differences between options. We were inspired to build Polarx to simplify this process. Our goal was to create a platform where users can easily compare policies side-by-side, helping them make smarter and faster decisions. The name Polarx derived from Polaris (the north star) reflects how like the north star we also help uers navigate complex choices.

What it does

Polarx is a platform designed to simplify how users understand and compare complex policies such as insurance plans and financial agreements. Instead of manually reading lengthy and technical documents, users can view multiple policies side-by-side in a clean, structured format. The platform highlights key differences, extracts important details, and uses AI to generate concise summaries, making dense policy language easier to understand. It also provides feasibility scores based on user conditions and an effectiveness rating to help users quickly evaluate how well a policy meets their needs. Additionally, users receive real-time alerts for important updates, ensuring they stay informed about changes that may impact them. For organizations, Polarx includes an admin dashboard that enables efficient management of policy data. Administrators can upload and update policy documents, track insurance types, manage associated drugs or coverage items, and monitor system usage. This dual functionality ensures that both end-users and organizations can interact with the platform effectively.

How we built it

We built Polarx using a full-stack architecture that combines modern web technologies, AI integration, and scalable database design to support both users and administrators. Frontend We used React to create a dynamic and responsive user interface. The UI was designed with Tailwind CSS and Material UI, allowing us to build clean, intuitive dashboards and comparison views. The frontend enables users to seamlessly compare policies, view AI-generated summaries, and track feasibility scores, effectiveness ratings, and real-time alerts.

Backend The backend was developed using Node.js with Express, which handles API requests, business logic, and user authentication. We implemented JWT-based authentication to ensure secure login and role-based access for both users and administrators. To power the core functionality, we integrated AI/ML APIs that process policy data, generate summaries, and evaluate coverage feasibility. The backend also manages notification services that trigger alerts when policies are updated or conditions change.

Database We used PostgreSQL (via Supabase) to store structured data, including user profiles, policy documents, insurance details, drug coverage data, and alerts. The database schema was designed to efficiently handle relationships between policies, users, and coverage elements.

Challenges we ran into

Parsing complex policy documents: Insurance and financial documents come in unstructured formats (PDFs, long text), making it difficult to extract meaningful and consistent data for comparison.
AI summarization accuracy: Ensuring that AI-generated summaries were both accurate and easy to understand without losing critical legal or financial details was challenging.
Designing meaningful comparisons: Deciding what actually matters to users (coverage, cost, exclusions, effectiveness) required careful thought and iteration.
Database structuring: Modeling relationships between policies, drugs, insurance types, and users in Supabase while keeping queries efficient was non-trivial.
Authentication & security: Implementing secure JWT-based authentication while managing roles (admin vs user/doctor) required careful backend design.
Real-time updates & alerts: Building a system that notifies users about policy changes or updates reliably added complexity.
Team coordination: Integrating frontend, backend, and AI components smoothly within limited time required strong collaboration.

Accomplishments that we're proud of

Built a fully functional full-stack platform from scratch within a short time
Successfully implemented side-by-side policy comparison UI, making complex data easy to understand
Integrated AI-powered summaries to simplify dense policy language
Designed a feasibility scoring system to help users understand coverage relevance
Created an admin dashboard for managing policies, drugs, and insurance data
Implemented secure authentication system (JWT)
Delivered a clean, intuitive UI/UX that non-technical users can easily navigate

What we learned

How to integrate AI into real-world applications beyond just basic prompts
The importance of data modeling when dealing with complex relational systems
How to design systems for both end-users and administrators
Practical experience with full-stack architecture (React + Node.js + PostgreSQL)
The challenge of balancing accuracy vs simplicity when presenting AI-generated insights
How to collaborate effectively across frontend, backend, and AI components
The importance of user-centric design when dealing with complex information