FinPilot

Inspiration

Most banking apps tell you where your money went, but what if they could also tell you where your finances are heading?

With rising living costs, unexpected expenses, and increasingly complicated financial decisions, we realized that simply tracking spending isn't enough. People need tools that help them understand their financial health, anticipate challenges, and make smarter decisions before problems arise.

Inspired by the PNC 360° Financial Health Navigator Challenge, we created FinPilot, an AI-powered personal financial intelligence platform designed to make financial planning more proactive, personalized, and accessible.

Our goal was simple: Build a financial copilot that doesn't just show users their money, but helps them understand what to do next.

What it does

FinPilot brings together financial analytics, predictive machine learning, and AI-powered guidance in one platform.

Its core features include:

Unified Financial Dashboard: View account balances, income, expenses, debt, net worth, and spending patterns in one place.
Financial Health Score: Understand overall financial wellness through an explainable 0–100 score based on cash-flow stability, emergency savings, debt management, spending habits, and payment readiness.
Predictive Cash Flow: Forecast upcoming spending and account balances using machine learning to identify potential financial challenges.
FinPilot Guardian: Receive proactive alerts about low projected balances, unusual spending, upcoming bills, and insufficient emergency savings.
Financial Digital Twin: Explore what-if scenarios such as losing income, reducing expenses, or facing an unexpected emergency, and see how those changes could affect financial stability.
Economic Intelligence: Understand how broader economic conditions, including inflation and interest rates, may affect personal finances.
AI Financial Copilot: Ask personalized questions about spending, financial health, upcoming expenses, and potential financial decisions.

What makes FinPilot different is how these features work together through our Detect → Simulate → Recommend approach.

For example, if Guardian detects that a user's checking balance might fall below a safe threshold before their next paycheck, the Digital Twin can simulate possible spending adjustments, while the AI Copilot explains the results and suggests potential next steps.

Rather than simply displaying financial information, FinPilot helps users turn insights into action.

How we built it

We developed FinPilot using a modular full-stack architecture, separating the user interface, financial data processing, machine learning, and AI-powered insights.

Frontend: We used Next.js, React, TypeScript, Tailwind CSS, and shadcn/ui to build a clean, responsive dashboard. Recharts powers our interactive visualizations for spending trends, financial health, and cash-flow forecasts.

Backend: We used Python and FastAPI to build REST APIs, process financial information, and connect the frontend with our financial intelligence engine. Supabase PostgreSQL provides structured storage for accounts, transactions, and financial records.

Financial Data: We worked with synthetic financial data and the PNC hackathon starter kit to develop and test our platform without requiring sensitive banking information.

Machine Learning: We used LightGBM, Pandas, NumPy, and scikit-learn to develop spending forecasts based on historical transaction patterns, rolling averages, and calendar-related features. We combined these predictions with expected income and recurring bills to estimate future cash flow.

Financial Intelligence: We developed deterministic financial calculations for health scoring, risk detection, and what-if simulations. This ensures that financial results come from defined calculations rather than relying on AI-generated numbers.

Generative AI: We integrated OpenAI's API with backend tool calling to power our financial copilot. The assistant retrieves relevant financial information, explains predictions, and helps users explore financial scenarios through natural-language conversations.

A major focus of our architecture was ensuring that forecasting, risk detection, simulations, and AI explanations all use consistent financial information.

Challenges we ran into

One of our biggest challenges was making FinPilot genuinely useful rather than building another dashboard with AI-generated summaries.

Ensuring financial accuracy: Financial calculations involve more than adding income and subtracting expenses. We had to consider recurring bills, debt obligations, transaction categories, and account balances while avoiding misleading or inconsistent results.

Working with synthetic data: Since we used simulated financial histories, we had to think carefully about realistic spending patterns, model evaluation, and the limitations of predictions made using synthetic data.

Connecting multiple systems: Our forecasting engine, financial health scoring, Guardian alerts, and Digital Twin simulations depend on one another. Ensuring these components produced consistent results required careful backend design.

Grounding AI responses: We wanted our copilot to answer personalized financial questions without inventing numbers. Connecting AI responses to actual backend calculations was an important technical challenge.

Managing hackathon time: With a limited development window, we had to balance functionality, technical accuracy, and user experience while prioritizing the features that made FinPilot valuable.

Accomplishments that we're proud of

We're particularly proud of building FinPilot around a connected financial decision-making experience rather than creating another traditional budgeting application.

Our biggest accomplishments include:

Combining financial analytics, predictive forecasting, and personalized financial insights into one platform.
Developing an explainable financial health scoring system that helps users understand their financial strengths and vulnerabilities.
Designing FinPilot Guardian to proactively identify potential financial problems before they occur.
Connecting the Digital Twin simulator with financial forecasts and risk detection so users can explore how different decisions might affect their future.
Integrating an AI Financial Copilot that uses backend tools to provide more personalized and grounded explanations.
Creating a user experience that makes complex financial information easier to understand and act upon.

Most importantly, we're proud that FinPilot focuses on helping users make better financial decisions, not simply presenting more financial data.

What we learned

Building FinPilot taught us that meaningful AI integration requires much more than adding a chatbot to an application.

We learned how predictive machine learning, deterministic financial calculations, and generative AI can complement one another when each is responsible for the right task.

We also gained a deeper understanding of financial data modeling, forecasting uncertainty, API integration, and the importance of maintaining consistency between different components of a complex application.

Working as a team helped us improve our ability to divide responsibilities, integrate independently developed features, and prioritize technical reliability under tight deadlines.

Our biggest takeaway was that AI is most valuable when it makes reliable information easier to understand and use, rather than replacing the underlying logic.

What's next for FinPilot

We see FinPilot as the beginning of a more intelligent and accessible approach to personal financial management.

Moving forward, we hope to:

Integrate real banking accounts through secure, permission-based financial data connections.
Improve forecasting accuracy by training and evaluating models on more diverse financial histories.
Expand Guardian with advanced anomaly detection, personalized thresholds, and proactive notifications.
Enhance the Digital Twin with more detailed scenarios involving debt repayment, long-term savings, and economic uncertainty.
Improve the AI Copilot with more personalized financial coaching and goal-based planning.
Introduce FinPilot Wrapped, an interactive weekly or monthly recap that makes financial insights more engaging and accessible.

Our long-term vision is to make FinPilot a financial intelligence companion that helps people confidently navigate everyday financial decisions and unexpected challenges.

FinPilot — Understand your money. Anticipate risks. Prepare for what's next.