Personal project

Coinpi - AI Financial Assistant

CoinPi - AI Financial Assistant

Compartir artículo

From an Excel sheet to a production React Native app. AI-powered financial assistant built with Expo, Supabase, and intelligent automation.

From Excel to Production Mobile App

“And to think that it all started from an Excel sheet…”

New year, new savings goals, new budgets… That’s how it began. A shared Excel template for weekly expense tracking became the perfect opportunity to finally dive into mobile development.

The origin story

When María Sugasaga shared her Excel template for expense control, I saw the opportunity to transform a spreadsheet into a native mobile app - and finally explore React Native development I’d been postponing.


Why React Native? Technical Decision-Making

After evaluating different mobile frameworks, React Native won for key technical advantages:

Development Efficiency

  • Single codebase for iOS and Android (~90% code sharing)
  • Hot Reloading for rapid iteration during development
  • Near-native performance through native component architecture

Ecosystem & Flexibility

  • Mature ecosystem of libraries and tools
  • JSI (JavaScript Interface) for low-level operations
  • Flexibility to integrate native modules for platform-specific optimizations

React to React Native

As a developer interested in React for web, jumping to React Native was a stimulating technical challenge. The transition leverages existing React knowledge while opening mobile development capabilities.


The Power of Expo & Supabase

Expo: Streamlined Mobile Development

We incorporated Expo as our build platform, which enabled us to:

  • Accelerate development without complex native configurations
  • Rapid testing using Expo Go during development
  • Pre-configured native APIs for common functionalities
  • Cloud builds optimized for both platforms

Supabase: Complete Backend Solution

For the backend, we chose Supabase as our all-in-one platform:

Authentication System

  • Robust and secure authentication implementation
  • Social login and email/password support

Edge Functions

  • Serverless functions for business logic
  • AI integration for expense analysis
  • Real-time data processing

PostgreSQL Database

  • All application data stored in production-grade infrastructure
  • Relational data modeling for transactions and categories
  • Built-in Row Level Security (RLS)

Real-Time Capabilities

  • Live data synchronization across devices
  • Instant updates without polling

Clean Architecture Implementation

Implemented a clean architecture with clear separation of concerns:

├── UI Layer (React Native Components)
├── Business Logic (Custom Hooks & Services)
└── Data Persistence (Supabase Client & Edge Functions)

This architecture ensures:

  • Maintainability as the codebase grows
  • Testability of business logic
  • Scalability for future features

Real User Impact

MetricImpactUser Feedback
Average monthly savings+20%“I finally see where my money goes”
Time managing finances-75%“No more manual Excel updates”
Expense visibility100%“Complete financial transparency”
Budget adherence+60%“Alerts keep me on track”

The real problem solved

Traditional finance apps show data but don’t help make decisions. CoinPi goes further: it analyzes, recommends, and learns from your habits using AI.


AI-Powered Intelligence

Personalized Recommendations

The AI analyzes spending patterns and provides:

  • Specific recommendations for your financial situation
  • Concrete actions based on real data, not generic advice
  • Behavioral insights to improve financial habits

Automatic Expense Analysis

  • Intelligent categorization of transactions
  • Learning from your patterns for automatic tagging
  • Anomaly detection for unusual expenses

Future Capabilities (In Development)

  • Predictive expense forecasting
  • Savings goals with automatic tracking
  • Smart alerts for unusual spending patterns
  • Banking integration for automatic transaction sync

Technical Stack Deep Dive

Frontend (Mobile)

  • Framework: React Native + Expo
  • Language: TypeScript
  • State Management: React Context + Custom Hooks
  • UI Components: Custom design system

Backend (Serverless)

  • Platform: Supabase
  • Functions: Edge Functions (Deno runtime)
  • API: RESTful + Real-time subscriptions
  • AI Integration: OpenAI GPT-4 for analysis

Database

  • Type: PostgreSQL (via Supabase)
  • ORM: Supabase Client Library
  • Security: Row Level Security (RLS)

DevOps

  • CI/CD: Expo Application Services (EAS)
  • Deployment: Automated builds for iOS & Android
  • Monitoring: Supabase Analytics + Custom logging

Next Technical Challenges

Secure Banking API Integration

  • Implementing PSD2-compliant banking connections
  • OAuth 2.0 for secure authorization
  • Transaction synchronization with Spanish banks

Performance Optimization

  • Bundle size reduction for faster app loads
  • Code splitting for lazy-loaded features
  • Image optimization and caching strategies

Advanced AI Features

  • ML models for spending prediction
  • Custom recommendation engine
  • Financial health scoring algorithm

Collaborative project

This project is being developed in collaboration with María Sugasaga as part of my technical exploration into mobile development, fintech, and AI integration.


Development Philosophy

This project embodies several key principles:

  • Agile development with iterative improvements
  • AI integration for intelligent user experiences
  • Continuous learning and skill expansion
  • Real-world problem solving over theoretical exercises

Connect with me

Interested in React Native, fintech, or AI integration? I’d love to connect with other developers to discuss technical approaches, optimizations, and best practices.

Current Status

  • Mobile app: Complete and in active use
  • Web version: In development
  • Banking integration: Planning phase
  • AI enhancements: Ongoing improvements
Alder Darío Velásquez Obando

Written by

Alder Darío Velásquez Obando

Full Stack Developer & DevOps Engineer passionate about technology, artificial intelligence and creating innovative solutions.

Martin

Hi! I'm Martin, Alder's Virtual assistant. How can I help you?