MoneyMatters

Finance Policy Tracker AI

Comment

Synaxis — DevPost Project Story

💡 Inspiration

Financial wellness is often treated as a privilege. If you can afford a financial advisor, you get a complete picture of your money — your spending, your savings, your risks, your coverage gaps. If you can't, you're left piecing it together alone.

We kept coming back to one uncomfortable truth: the tools that exist today only solve half the problem. Budgeting apps like Mint or YNAB tell you where your money went. Insurance platforms like Policygenius help you buy coverage. But no one had built a single place that connects the two — that looks at your full financial life and says "here's what you have, here's what you're missing, and here's what happens if life goes sideways."

For millions of Americans — first-generation immigrants, young earners, gig workers, people in underserved communities — that gap isn't just inconvenient. It's dangerous. A single unexpected medical bill or job loss can wipe out years of savings when you never knew you were underinsured in the first place.

That's what inspired Synaxis. We wanted to build the tool we wish existed — one that treats financial wellness as a right, not a luxury.

🔨 How We Built It

Synaxis is a full-stack mobile application built with a clean separation between the frontend experience and the backend intelligence layer.

Frontend — We built the mobile app in React Native with Expo, using TypeScript throughout for type safety and maintainability. The UI was designed mobile-first with a focus on accessibility and clarity — clean card-based layouts, intuitive navigation, and visual data representations that don't require a finance degree to understand.

Backend — The API layer runs on Node.js with Express, handling user authentication via JWT tokens, transaction management, and routing requests to the AI engine. We structured the backend to be modular so the Financial Tracker and Synaxis AI modules could evolve independently.

AI Integration — The Synaxis AI module is powered by the Claude API (Anthropic). When a user uploads an insurance policy document, we pass it to Claude with a structured prompt that instructs it to identify coverage gaps, flag underinsured areas, and return findings in plain, jargon-free language. The same AI layer powers our scenario simulation — users describe a "what-if" situation and Claude generates a realistic financial impact analysis based on their actual data.

Infrastructure — The entire application is containerized using Docker and Docker Compose with separate configurations for development and production environments. We set up a GitHub Actions CI/CD pipeline so deployments are automated on every push to main.

🚧 Challenges We Faced

1. Prompting the AI for Consistent, Structured Output Getting Claude to return insurance gap analysis in a format that was both human-readable and parseable by our frontend took significant iteration. Insurance documents are dense, inconsistently formatted, and full of legal language. We had to carefully engineer our prompts to handle edge cases — missing sections, ambiguous clauses, and policies with overlapping coverage — without hallucinating gaps that didn't exist.

2. Making Complex Data Feel Simple Financial data is inherently noisy. Displaying 48 months of income and expense trends in a way that's instantly readable on a small mobile screen — without overwhelming the user — required multiple rounds of UI iteration. We rebuilt the chart components several times before landing on something that felt clean and actionable rather than cluttered.

3. Unifying Two Very Different Product Surfaces The Financial Tracker and Synaxis AI are fundamentally different tools with different data models, different interaction patterns, and different user mental models. Designing a home dashboard that made both feel like part of one cohesive product — rather than two apps stapled together — was one of our biggest UX challenges.

4. Building for Real-World Edge Cases Real user finances are messy. Irregular income, multiple expense categories, months with no transactions, insurance policies with exclusion clauses — we had to make sure the app didn't break or produce misleading outputs when the data wasn't clean and predictable.

📚 What We Learned

  • AI is only as useful as the context you give it. The quality of our insurance analysis improved dramatically once we started feeding Claude structured context alongside the raw document — user income, existing coverage, family size — rather than just dumping the PDF.

  • Accessibility is a design constraint, not an afterthought. Building for underserved communities meant questioning every default assumption: Is this label clear to someone who didn't grow up speaking financial jargon? Does this work on a low-end Android device? Would a non-native English speaker understand this output?

  • The hardest part of a full-stack hackathon isn't the code — it's staying aligned. With both a complex frontend and a backend + AI layer running in parallel, we learned quickly that clear API contracts and regular sync-ups mattered more than raw coding speed.

  • Docker saves you. Setting up containerized environments early meant we spent zero time on "works on my machine" issues during the final stretch.

🔮 What's Next for Synaxis

  • 📍 Location Services — Help users find nearby financial advisors, insurance brokers, and community financial resources filtered by specialty and proximity, with a focus on surfacing professionals who serve underrepresented communities.

  • 🤖 Proactive AI Suggestions — Instead of waiting for users to ask questions, Synaxis will proactively analyze spending patterns and coverage data to send personalized alerts: "Your emergency fund covers 2 months of expenses — industry standard is 6. Here's a plan to get there."

  • 🌐 Multi-Language Support — Full localization for Spanish, Mandarin, Hindi, and other languages spoken by communities most underserved by current financial tools.

  • 🤝 Insurance Carrier API Integration — Real-time policy quotes from carriers directly inside the app, so users can go from "I have a gap" to "I'm covered" without leaving Synaxis.

Share this project:

Updates