🌸 Sahej — Smart Investing for Every Woman

Sahej is an AI-powered financial advisor built for Indian housewives who save regularly but lack access to personalised investment guidance. It combines a multi-agent CrewAI system with a clean chat interface to make investing simple, trustworthy, and accessible.

---

Features

• Google SSO — one-tap sign-in, no passwords
• AI-powered advice — multi-agent CrewAI crew (Financial Advisor, Investment Researcher, Savings Coach) answers questions in plain Hindi or English
• Live investment data — fetches current rates for PPF, NSC, Sukanya Samriddhi, FDs, SIPs, and government schemes via web search
• SIP calculator — projects returns for any monthly investment amount and duration
• Chat history — all conversations persisted per user in Supabase
• Built for India — understands Indian investment context, government schemes, and household savings patterns

---

Tech Stack

Layer	Technology
Frontend	React + Vite + TypeScript
Backend	FastAPI (Python)
AI	CrewAI 1.x + OpenAI GPT-4o-mini
Auth	Supabase Auth (Google OAuth)
Database	Supabase (PostgreSQL)
Search	Serper API (Google Search)

---

Project Structure

sahej/
├── backend/
│   ├── app/
│   │   ├── api/routes/       # FastAPI routes (chat)
│   │   ├── core/             # Config and settings
│   │   ├── crew/
│   │   │   ├── agents/       # BaseAgent + FinancialAdvisor, Researcher, Coach
│   │   │   ├── tools/        # SIPCalculatorTool, SchemeLookupTool
│   │   │   ├── llm.py        # LLMProvider singleton
│   │   │   └── sahej_crew.py # Crew orchestration
│   │   ├── schemas/          # Pydantic schemas
│   │   └── services/         # Supabase HTTP client
│   ├── requirements.txt
│   └── .env.example
├── frontend/
│   ├── src/
│   │   ├── api/              # Axios client + chat API
│   │   ├── lib/              # Supabase JS client
│   │   └── pages/            # LoginPage, ChatPage
│   └── .env.example
├── supabase_tables.sql       # Run in Supabase SQL editor
├── PRD.md
└── USER_STORIES.md

---

Getting Started

Prerequisites

• Python 3.12+
• Node.js 18+
• Supabase account
• OpenAI API key
• Serper API key (serper.dev)

1. Supabase setup

1. Create a Supabase project
2. Run supabase_tables.sql in the SQL editor
3. Enable Google provider: Authentication → Providers → Google

2. Backend

cd backend
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

cp .env.example .env
# Fill in SUPABASE_URL, SUPABASE_ANON_KEY, OPENAI_API_KEY, SERPER_API_KEY

uvicorn app.main:app --reload

3. Frontend

cd frontend
npm install

cp .env.example .env
# Fill in VITE_SUPABASE_URL, VITE_SUPABASE_KEY

npm run dev

Open http://localhost:5173

---

Architecture — CrewAI Agents

User message
     │
     ▼
┌─────────────────────┐
│   Savings Coach     │  identifies goals, time horizons, savings capacity
└────────┬────────────┘
         │ context
         ▼
┌─────────────────────┐
│ Investment Researcher│  searches live rates, finds matching schemes + products
└────────┬────────────┘
         │ context
         ▼
┌─────────────────────┐
│  Financial Advisor  │  writes the final warm, actionable reply to the user
└────────┬────────────┘
         │
   Response saved to Supabase → returned to frontend

All three agents share a single LLM instance (LLMProvider singleton) using OOP polymorphism — each subclasses BaseAgent and overrides role, goal, backstory, and tools.

---

Environment Variables

Backend .env

Variable	Description
SUPABASE_URL	Your Supabase project URL
SUPABASE_ANON_KEY	Supabase anon/publishable key
OPENAI_API_KEY	OpenAI API key
SERPER_API_KEY	Serper API key for live search
CORS_ORIGINS	JSON array, e.g. ["http://localhost:5173"]

Frontend .env

Variable	Description
VITE_SUPABASE_URL	Your Supabase project URL
VITE_SUPABASE_KEY	Supabase anon key
VITE_API_URL	Backend URL (default: http://localhost:8000/api)