North Star

The budgeting page
The dashboard
The investment page

Inspiration

We were inspired by the complexity and anxiety many people face when managing their personal finances. Traditional tools are often clunky, filled with jargon, or lack personalized guidance. We wanted to build an application that simplifies personal finance, making intelligent budgeting and sophisticated investment strategies accessible to everyone, regardless of their financial literacy. Our goal was to create a co-pilot that demystifies money management through the power of AI.

What it does

NorthStar is an AI assisted personal finance application designed to guide users toward financial wellness. A new user begins with a simple onboarding questionnaire about their age, income, and financial situation. The app instantly provides a personalized recommendation for splitting their income between budgeting and investing. From the main dashboard, users can dive into two core areas. The budgeting section provides a detailed breakdown of spending, analyzes recurring subscriptions for potential savings, and uses AI to generate actionable insights to optimize their cash flow. The investing section helps users determine their risk tolerance, presents three tailored investment portfolios (Conservative, Moderate, and Aggressive), and allows them to view and understand their portfolio composition with the help of an AI copilot that can answer questions in plain language.

How we built it

We built NorthStar as a modern, full stack application using Next.js with the App Router, which allowed us to seamlessly integrate the frontend and backend. The user interface is built with React and TypeScript for type safety, and styled using Tailwind CSS with the shadcn/ui component library for a clean and responsive design. Data visualizations are powered by Recharts. The backend logic is handled through Next.js API Routes running on a Node.js environment. For our MVP, we engineered a stateless architecture using a session based, in memory mock data store, which avoids the need for a database and allows for a lightweight, fast user experience. All AI features, including budget insights and the investment chatbot, are powered by calls to large language models like GPT 4, which are handled securely on the server side. The entire application is designed for deployment on Vercel.

Challenges we ran into

One of the biggest challenges was engineering the AI integration to provide consistently useful and safe financial guidance. This required extensive prompt engineering to ensure the AI generated structured, actionable insights without giving prescriptive or dangerous advice. Another significant challenge was designing a robust and interactive user experience within the constraints of a stateless MVP. Managing application state without a persistent database meant carefully architecting a combination of React Query, Context, and localStorage to create a feeling of continuity for the user throughout their session. Finally, designing the comprehensive data models for profiles, budgets, and investment portfolios from scratch was a complex task that required a lot of foresight.

Accomplishments that we're proud of

We are incredibly proud of creating a comprehensive and polished user experience that feels like a real, data driven product despite its stateless architecture. The seamless integration of AI throughout the application is a major accomplishment; it acts as a helpful guide rather than a gimmick. We are also proud of the detailed and robust system design itself, including the versioned API, the structured component hierarchy, and the thorough testing strategy. Building a feature rich application that can run entirely in memory for demo purposes is a technical achievement that allowed us to iterate and deploy very quickly.

What we learned

Throughout this project, we learned a great deal about the practical application of large language models. We gained deep experience in prompt engineering and learned how to reliably parse structured JSON output from an AI to power a user interface. Building NorthStar solidified our understanding of modern full stack development with Next.js, highlighting the power of a single framework for both client and server logic. We also learned the importance of a well defined state management strategy, especially how to effectively combine server state tools like React Query with client state solutions to build complex, responsive applications.

What's next for North Star

The current version of NorthStar is a powerful proof of concept, and we have a clear roadmap for the future. The next major step is to move beyond the MVP's stateless architecture by introducing a persistent database like PostgreSQL to store user data and a Redis cache for session management. Following that, we plan to integrate with third party services like Plaid to allow users to connect their real bank accounts for live transaction analysis. We also intend to build out a full user authentication system to enable data persistence and cross device synchronization. Long term, we will explore connections to broker APIs to facilitate real trading and expand our AI capabilities with more predictive financial modeling.