🌱 Sprout-Sell — AI-Powered Credit For Every Farmer

Bold, inclusive, and data-driven. Sprout-Sell brings AI credit assessments to smallholder farmers using simple inputs (name, location, land size), and enriches the analysis with climate, weather, and yield signals — no financial history required.

• Frontend (Netlify): https://mazao-credit.netlify.app
• Backend (Render): https://mazao-credit-backend.onrender.com

---

🧠 How the AI Works

Our AI computes a creditworthiness assessment by blending farmer-provided inputs with external and modelled datasets:

• Land and production context: land size, crop type, years of experience
• Location awareness: county/region, country, regional yield baselines
• Climate and weather: risk of drought, flood, and hail by crop and county
• Optional mobile money aggregates: inflows, outflows, transaction counts

The pipeline can run entirely without sensitive bank statements. Financial data, if provided, simply refines the baseline. The AI returns:

• Credit score (0–100)
• Recommended loan amount
• Interest rate
• Strengths, weaknesses, and risk level
• Financial readiness and analysis timestamp

---

📦 Project Structure

Sprout-Sell/
  Sprout-Sell/
    ai/
      data/                    # CSV inputs (e.g., weather_forecast.csv)
      scripts/                 # Python AI entrypoint
        credit_predictor.py
      models/                  # Optional model artifacts (.pkl)
      venv/                    # Python virtual environment
    backend/
      src/
        app.js                 # Express server
        routes/
          auth.js              # Registration + credit analysis API
          financial.js
          products.js
          users.js
        services/
          AICreditScoring.js   # JS/OpenAI fallback + parsing
          PythonCreditScoring.js# Spawns ai/scripts/credit_predictor.py
        config/
          supabase.js          # Supabase client (service key on server)
    frontend/
      src/
        components/
          FinancialTools.tsx   # Minimal input → AI analysis + raw JSON view
        pages/
          Auth.tsx             # Signup (with gender) + role-aware
          FinancialPage.tsx
          Dashboard.tsx
      index.html
      vite.config.ts
    netlify.toml                # SPA build + redirects
    render.yaml                 # Backend deploy config
    README.md

---

🔗 Key Endpoints

• POST /api/auth/register — Create/update user profile (includes role, gender, data)
• POST /api/auth/credit-analysis/:userId — Run AI; persists score and logs to server
• GET /api/auth/credit-analysis/:userId — Fetch latest persisted analysis

The backend first tries Python AI (ai/scripts/credit_predictor.py); if unavailable, it falls back to a JS/OpenAI model, then persists the result to Supabase (when configured).

---

🚀 Run Locally

1) AI Service (Python)

cd Sprout-Sell/ai
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt  # if present; otherwise install numpy, pandas, etc.

# Optional: seed weather data
echo "county,crop,drought_risk,flood_risk,hail_risk\nnakuru,maize,0.2,0.1,0.05\nnairobi,beans,0.1,0.1,0.02\nmachakos,tomatoes,0.3,0.05,0.03" > data/weather_forecast.csv

# Smoke test
echo '{
  "farmData":{"farmSize":3,"farmType":"maize"},
  "locationData":{"county":"Nakuru"},
  "financialData":{"annualRevenue":120000},
  "mpesaData":{"total_inflows":80000,"total_outflows":40000,"inflow_count":120}
}' | python scripts/credit_predictor.py

2) Backend (Node/Express)

cd Sprout-Sell/backend
npm install

# Optional persistence (recommended)
export SUPABASE_URL=your_supabase_url
export SUPABASE_SERVICE_KEY=your_service_role_key

# Start API
node src/app.js
# → http://localhost:5000

3) Frontend (Vite + React)

```bash

cd Sprout-Sell/frontend npm install

Create .env.local

cat > .env.local <<EOF VITE_API_URL=http://localhost:5000 VITE_SUPABASE_URL=your_supabase_url VITE_SUPABASE_ANON_KEY=your_supabase_anon_key EOF

npm run dev

→ http://localhost:5173

---

## 🧪 Quick API Test
```bash
# Replace with a real Supabase user id once signed up via frontend
USER_ID="test-user-123"

curl -s -X POST "http://localhost:5000/api/auth/credit-analysis/$USER_ID" \
  -H "Content-Type: application/json" \
  -d '{
        "farmData": {"farmSize": 2.5, "farmType": "beans", "yearsExperience": 2},
        "locationData": {"region": "Nairobi", "country": "Kenya"},
        "financialData": {"annualRevenue": 20000},
        "mpesaData": {"total_inflows": 15000, "total_outflows": 12000, "inflow_count": 40}
      }'

The server logs a line like: [AI CREDIT] user=... score=... rate=... loan=... and persists to Supabase if configured.

---

🌍 Deployments

Frontend (Netlify): https://mazao-credit.netlify.app
Backend (Render): https://mazao-credit-backend.onrender.com

Netlify build (configured in netlify.toml):

• Base: frontend
• Build: npm run build
• Publish: dist
• Env: VITE_API_URL, VITE_SUPABASE_URL, VITE_SUPABASE_ANON_KEY

---

✨ Why This Matters

Millions of farmers lack formal financial records. Sprout-Sell flips the script by leveraging agronomic signals — yield baselines, climate risk, location, land size, and experience — to open fair credit access. Finance should follow good farming, not just bank statements.