DollarPilot

Inspiration

Traditional personal finance apps are built for accountants. They rely on intimidating spreadsheets, complex jargon, and boring dashboards that alienate everyday users—especially Gen-Z and Millennials. We realized that people don't want to be lectured; they want to be engaged. We wanted to build an app that felt less like a bank and more like a brutally honest, highly intelligent friend. That’s how DollarPilot was born: a unified financial companion that combines empathy, data visualization, and a little bit of tough love.

What it does DollarPilot is a 3-in-1, AI-powered personal finance Single Page Application (SPA). It features three core pillars:

The Finance Coach: A context-aware conversational AI that acts as a jargon-free personal advisor, helping you budget and understand complex financial concepts.

The What-If Predictor: A financial time machine. Users input scenarios (e.g., "What if I invest $500/month instead of partying?"), and the AI instantly generates the math, rendering a beautiful, interactive compound growth chart.

Roast Your Finance & Fix: A viral "Spotify Wrapped" for bad spending habits. Users input their guilty spending via interactive sliders, select an AI persona (like a Wall Street Bro or a Disappointed Mafia Boss), and receive a hilarious, shareable "Red Flag" scorecard followed by serious, actionable steps to fix the bleeding.

How we built it We focused on speed and a gorgeous UI.

Frontend & UI: We built a lightning-fast SPA using Next.js (App Router), styled with Tailwind CSS and shadcn/ui for the sleek tabs, cards, and interactive sliders.

AI Engine: We powered the logic using the Anthropic Claude 3 API, utilizing strict system prompts to maintain the personas and generate context-aware advice.

Data Visualization: For the Predictor, we engineered our Claude prompts to return strictly formatted JSON, which we fed directly into Recharts to dynamically render the growth charts in real-time.

Viral Loop: We integrated html-to-image to let users download their Roast Scorecards directly to their devices for social media sharing.

Challenges we ran into The biggest technical hurdle was getting an LLM to reliably generate data for our charts. Claude is great at text, but forcing it to return a perfect, parse able JSON object—containing both a narrative string and an array of year-by-year data points—required heavy prompt engineering. If the JSON was malformed even slightly, it would break the Recharts component. We also spent time fine-tuning the "Roast" personas to ensure they were funny and savage without crossing the line into being unhelpful or offensive.

Accomplishments that we're proud of We are incredibly proud of the What-If Predictor's dynamic charting. Seeing an AI take a natural language scenario, calculate the compound interest, and seamlessly render an animated Recharts graph all in a matter of seconds is a massive "wow" moment. We're also proud of executing a highly polished, unified UI within a tight 6-hour hackathon window, proving that complex AI tools don't need clunky interfaces.

What we learned We learned the sheer power of Structured Outputs (JSON) from LLMs. Moving beyond simple chat interfaces and using AI as a backend data processor opens up massive possibilities for web apps. We also learned a valuable product lesson: humor and gamification (the Roast feature) are incredibly effective ways to get users to engage with otherwise stressful topics like personal debt and budgeting.

What's next for DollarPilot Plaid / Account Integration: Moving from interactive sliders to pulling real transaction data to automate the Roast and Coaching context.

Community Leaderboards: Allowing users to anonymously share their "Red Flag Scores" to see how their spending habits compare to their peers.

More Visualizations: Expanding the Predictor to handle debt payoff timelines, mortgage amortizations, and side-hustle tax calculations.