Prisma

Unify. Analyze. Optimize. Your complete financial command center.

Comment

About Prisma

What Inspired This Project

Portfolio fragmentation is a real problem I've personally experienced. Managing investments across multiple platforms—Robinhood for stocks, Coinbase for crypto, real estate properties, and various bank accounts—made it impossible to see my complete financial picture. I was constantly switching between apps, manually tracking assets in spreadsheets, and never truly understanding my portfolio's risk exposure or diversification.

The inspiration came from frustration: Why isn't there a single platform that unifies everything and provides intelligent insights?

Most existing solutions either:

  • Only support one asset class (stocks OR crypto, not both)
  • Require expensive advisory services
  • Lack AI-powered analysis
  • Don't offer real-time stress testing

I wanted to build something that solves this comprehensively.

What I Learned

Building Prisma taught me several valuable lessons:

Technical Skills:

  • Universal API Integration: Learned to work with multiple financial APIs (SnapTrade, Alpha Vantage, Clerk) and handle their different authentication methods, rate limits, and data formats
  • AI Integration: Discovered how to effectively prompt and structure requests to Claude AI for financial analysis, ensuring reliable JSON responses and meaningful insights
  • Subscription Monetization: Gained deep understanding of RevenueCat's architecture, entitlements, offerings, and how to properly gate features server-side
  • Real-Time Data Synchronization: Built efficient systems to sync portfolio data from multiple sources while maintaining data consistency

Product Insights:

  • User Experience Matters: Financial apps need to be both powerful and approachable. Complex metrics like VaR need clear explanations
  • Security is Non-Negotiable: When handling financial data, every endpoint must be protected, every API call validated, and user privacy maintained
  • Performance is Critical: Portfolio calculations with 100+ assets need to be fast. Optimized database queries and efficient algorithms make the difference

Business Understanding:

  • Pricing Strategy: Learned to balance feature value with market expectations. The Pro tier at $14.99/month competes with YNAB while offering unique AI features
  • Market Positioning: Identified the gap between basic trackers (Mint) and expensive professional tools ($100+/month)

How I Built It

Architecture Decisions

I chose React Native with Expo for cross-platform development, allowing me to build for both iOS and Android simultaneously. The modular architecture separates concerns:

  • Frontend: React Native screens with NativeWind for styling, Clerk for authentication, RevenueCat for subscriptions
  • Backend: Node.js + Express API that handles all business logic, AI integration, and data aggregation
  • Database: MySQL for persistent storage of manual assets and user properties
  • External Services: Integrated SnapTrade (brokerage), Alpha Vantage (market data), Claude AI (analysis), Clerk (auth), RevenueCat (payments)

Key Features Built

  1. Universal Asset Aggregation

    • SnapTrade integration for 20+ brokerages
    • Manual entry system for real estate, crypto, commodities
    • Real-time price updates from Alpha Vantage

  2. AI-Powered Analysis

    • 15+ risk metrics (VaR, Beta, Sharpe, Sortino, Max Drawdown)
    • Stress testing against real economic scenarios
    • Custom scenario builder with Claude AI explanations
    • IPO alerts from market news

  3. Subscription System

    • RevenueCat integration with Pro Monthly ($14.99) and Pro Yearly ($149)
    • Server-side feature gating
    • Seamless purchase and restore flows

  4. Enhanced Dashboard

    • Portfolio insights (diversification score, risk level, top holdings)
    • Market news feed filtered by holdings
    • Clean, modern UI

Development Process

The project followed an iterative approach:

  1. MVP Core: Authentication, basic asset tracking, simple dashboard
  2. Analysis Layer: Risk metrics, correlation matrix, basic stress testing
  3. AI Integration: Claude API for scenario analysis and explanations
  4. Monetization: RevenueCat subscription system with proper feature gating
  5. Polish: UI improvements, error handling, performance optimization

Challenges Faced

Technical Challenges

1. API Rate Limiting & Data Consistency

  • Multiple APIs with different rate limits required careful request management
  • Solution: Implemented caching strategies and request queuing

2. AI Response Reliability

  • Claude sometimes returned malformed JSON or markdown-wrapped responses
  • Solution: Built robust JSON parsing with fallbacks, cleaned numeric formatting (removed commas), and extracted JSON from code blocks

3. Real-Time Portfolio Calculations

  • Calculating 15+ risk metrics for large portfolios (100+ assets) needed to be fast
  • Solution: Optimized algorithms, database indexing, and memoization of expensive calculations

4. Subscription State Management

  • Ensuring subscription status is always accurate across app restarts and network issues
  • Solution: Server-side validation on critical operations, local caching with periodic refresh

5. Android Permissions

  • Unnecessary permissions (like CAMERA) were being requested, requiring privacy policy
  • Solution: Removed unused dependencies (expo-image-picker) and explicitly set empty permissions array

Product Challenges

1. Making Complex Metrics Understandable

  • Users don't understand VaR or Sharpe Ratio without context
  • Solution: Added clear explanations, color-coded risk levels, and plain-English AI explanations

2. Balancing Feature Access

  • Deciding what should be free vs. paid
  • Solution: Made core tracking free, gated advanced AI features and unlimited accounts behind Pro subscription

3. User Experience for Financial Data

  • Financial apps can feel overwhelming
  • Solution: Clean UI, progressive disclosure (tabs for Overview, Risk, Opportunities), and actionable insights rather than raw data dumps

What's Next

If this project continues, I'd focus on:

  1. Enhanced Integrations: Plaid for US banks, Coinbase OAuth, more brokerages
  2. Advanced Features: Tax loss harvesting, rebalancing recommendations, goal-based planning
  3. Performance: Further optimization, offline support, faster sync times
  4. User Feedback: Iterate based on real user needs and pain points

Final Thoughts

Building Prisma was both challenging and rewarding. It pushed me to learn new technologies, think deeply about user experience, and solve real problems I've personally faced. The combination of universal aggregation, AI-powered insights, and stress testing creates something unique in the personal finance space.

The project demonstrates that with the right tools (React Native, modern APIs, AI), a solo developer can build a production-ready financial platform that rivals much larger teams.

Built with React Native, Node.js, Claude AI, and a lot of coffee.

Share this project:

Updates