FinWise — Coach de Educação Financeira com IA
aaravjj2/finwise
Coach IA de finanças para os desbancarizados, em 15 idiomas
Ver no GitHubFinWise é um coach de educação financeira alimentado por IA projetado para os 1,4 bilhão de adultos desbancarizados em mercados emergentes. Oferece tutoria conversacional em 15 idiomas, detector de fraudes, comparador de remessas e rastreador de gastos em uma PWA offline-first que funciona em celulares básicos. Combina streaming de IA, base de dados de instituições microfinanceiras verificadas e 30 lições interativas para alfabetizar financeiramente populações sem acesso bancário tradicional.
Instalação
npx convez-ai aaravjj2/finwise1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Educação financeira para população desbancarizada em mercados emergentes
- Detecção e análise de risco de ofertas de crédito predatório
- Comparação de taxas de remessas internacionais com dados reais
- Rastreamento de renda/despesa e planejamento de metas de poupança para microempreendedores
Veredito

Projeto bem pensado e tecnicamente sólido que resolve um problema real com alcance global. A arquitetura offline-first, suporte multilíngue nativo e ausência de fricção de signup são diferenciais reais — mas o modelo de IA é atualmente mock/Ollama, o que limita a sofisticação da tutoria em produção. Recomendado para quem quer incluir educação financeira de baixa renda em produtos existentes.
README
A tradução para o português deste README ainda está na fila.
🌍 FinWise — AI Financial Coach for Emerging Markets
Maya speaks your language, knows your market, and is completely free — AI financial coaching for the 1.4 billion unbanked adults in emerging markets.
🌐 Live Demo
Try FinWise now → https://finwise-azure.vercel.app
No signup required. Click "Continue without account" to explore the full product instantly.
The Problem
Amara is 32. She sells vegetables at a market stall in Lagos. Last month someone offered her a loan at "5% per week — very reasonable!" She almost accepted. That's 260% annual interest. She didn't know.
1.4 billion adults have no bank account. Without financial literacy:
- They pay 300%+ APR to loan sharks because they don't know alternatives exist
- They lose 6–10% of every remittance to fees 10× higher than necessary
- They can't build credit, access insurance, or start businesses safely
The Solution
FinWise is Maya — an AI financial coach who meets users where they are: on a $50 Android phone, in their language, explaining money in terms that make sense to their daily life.
✨ Features
| Feature | Description |
|---|---|
| 🤖 AI Coach (Maya) | Streaming conversational coaching in 15 languages |
| 🛡️ Scam Detector | Paste any offer — get instant risk analysis with APR calculation |
| 🏦 Institution Finder | 10+ verified microfinance institutions across 8 countries |
| 💸 Remittance Comparison | Real fee data for 5 major money transfer corridors |
| 🎓 Learning Modules | 6 modules, 30 lessons, quizzes, badges, progress tracking |
| 💰 Financial Dashboard | Income/expense tracking, savings goals, health score |
| 📱 Offline PWA | Installs to home screen, works without internet |
| 🔒 Demo Mode | Full product access, zero signup friction |
🌍 Languages
English · हिन्दी · Kiswahili · Yorùbá · বাংলা · Filipino · Español · Português · Hausa · አማርኛ · isiZulu · Igbo · தமிழ் · Bahasa Indonesia · Tiếng Việt
🛠️ Tech Stack
Frontend: Next.js 14 · TypeScript · Tailwind CSS · Zustand · next-intl Backend: Supabase (PostgreSQL + Auth + RLS) · Vercel Edge AI: Mock AI layer → Ollama (llama3.2:3b) → Claude API Testing: Vitest (45 tests) · Playwright (30 E2E) · GitHub Actions PWA: Service Worker · IndexedDB offline sync
🚀 Quick Start
git clone https://github.com/aaravjj2/finwise.git
cd finwise
npm install
cp .env.example .env.local
# Edit .env.local with your Supabase credentials
# ANTHROPIC_API_KEY=mock ← works without any API key
npm run dev
With local AI (free, no API key):
curl -fsSL https://ollama.ai/install.sh | sh
ollama pull llama3.2:3b
# Add to .env.local: OLLAMA_BASE_URL=http://localhost:11434
npm run dev
📊 Architecture
User (Android PWA)
↓ HTTPS
Vercel Edge (Next.js 14)
├── /api/chat → AI streaming (Ollama/Mock)
├── /api/health → Status endpoint
└── /api/scam-analyze → Risk assessment
↓
Supabase PostgreSQL
├── users, conversations, messages
├── learning_modules, lessons, user_progress
├── financial_entries, savings_goals
└── institutions, remittance_providers
🗺️ Roadmap
- Now: Mock AI, 15 languages, 6 modules, scam detector, institution finder
- Month 2: Ollama local AI on $20/month VPS (zero API costs)
- Month 5: Lightpanda web crawler — live institution rates, automated nightly
- Month 10: B2B API for MFI integration
- Month 16: Savings circles (chama/susu/tanda digitization)
- Month 21: On-device WASM inference — fully offline AI
📄 License
MIT — see LICENSE
🙏 Built For
The 1.4 billion people who deserve the same financial guidance that used to require a banker, a lawyer, and a financial planner.