मुनीम · Munim

AI Business Intelligence for Indian Small Businesses

Upload any financial file. Get instant AI analysis. Receive WhatsApp reports in Hindi or English. Built for Indian SMBs — kirana stores, medical shops, textile traders, and every business in between.

What It Does

📤  Upload Tally XML / Excel / CSV
        ↓
🤖  Gemini AI detects your column names automatically — no setup needed
        ↓
📊  Analytics engine computes revenue, products, customers, anomalies
        ↓
✨  Gemini generates 4 business-specific AI insights
        ↓
💬  WhatsApp report sent in Hindi / English / Hinglish

For Business Owners	For CA Firms
Upload Tally export → instant dashboard	Manage entire client portfolio
Revenue trends, top products, dead stock	At-risk client alerts
RFM customer segmentation	Bulk report generation
Anomaly detection (drops, spikes)	Per-client drill-down views
WhatsApp reports in your language	Client-level analytics

Architecture

┌──────────────────────────────────────────────────────────────────┐
│                        CLIENT LAYER                               │
│   📱 WhatsApp (Twilio / Meta)   🌐 Next.js 15 (App Router)       │
└────────────────┬─────────────────────────┬───────────────────────┘
                 │                         │
                 ▼                         ▼
┌──────────────────────────────────────────────────────────────────┐
│                     API LAYER  (FastAPI)                          │
│   /upload  /analysis  /reports  /auth  /ca  /whatsapp            │
└───────────────┬─────────────────────────┬────────────────────────┘
                │                         │
       ┌────────▼──────────┐    ┌─────────▼──────────┐
       │    Ingestor        │    │     Analytics       │
       │  ───────────────  │    │  ─────────────────  │
       │  Tally XML        │    │  metrics.py         │
       │  Excel / XLSX     │    │  anomaly.py         │
       │  CSV              │    │  rfm.py             │
       │  Gemini schema    │    │  seasonality.py     │
       │  detection        │    │  ai_insights.py     │
       └────────┬──────────┘    └─────────┬──────────┘
                │                         │
                └──────────┬──────────────┘
                           ▼
               ┌───────────────────────┐
               │    Neon PostgreSQL    │
               │  ─────────────────── │
               │  users               │
               │  uploads             │
               │  analysis_results    │
               │  reports             │
               └───────────────────────┘

Features

🤖 AI-Powered Schema Detection

Upload any financial file — Munim uses Gemini to understand your column names automatically. No configuration, no column mapping. Works with:

Bill Date, Invoice Date, Transaction Date, Date, तारीख
Net Total, Amount, Total, Sale Value
Party Name, Customer, Client, Buyer
Separate Debit / Credit columns from bank statements

📊 Complete Business Analytics

Revenue Metrics — Current vs previous period with % change and trend direction
Top Products — Ranked by revenue with trend indicators
Dead Stock Detection — Products with no sales in 14+ days
Customer Segmentation — RFM analysis: Champions, Loyal, At Risk, Lost
Anomaly Detection — Revenue drops, unusual spikes, pattern breaks
Adaptive Periods — Auto-detects if data spans a week, month, or quarter

✨ AI Business Insights

After every analysis, Gemini generates 4 actionable insights specific to your business type:

🟢 Opportunity — Growth levers to act on
🟡 Warning — Issues that need attention
🟠 Celebration — Wins worth acknowledging
🔵 Action — Specific things to do this week

💬 WhatsApp Reports

Narrative reports in Hindi, English, or Hinglish
Sent via Twilio sandbox (no business verification needed) or Meta Business API
Written by Gemini like a trusted advisor, not a robot
Template fallback if AI is unavailable — pipeline never fails

🏪 Supported File Formats

Format	Description
Tally XML	Export from TallyPrime — Gateway of Tally
Excel (.xlsx, .xls)	Any spreadsheet with date + amount columns
CSV	Any comma-separated file

Tech Stack

Layer	Technology
Frontend	Next.js 15, TypeScript, Tailwind CSS, Recharts
Backend	FastAPI, Python 3.12, SQLAlchemy (async)
Database	Neon PostgreSQL (serverless)
AI	Google Gemini Flash (`google-genai` SDK)
Auth	NextAuth.js v5, Google OAuth
WhatsApp	Twilio (primary) + Meta Business API (fallback)
File Parsing	openpyxl, pandas, lxml
Testing	pytest — 122 tests

Quick Start

Prerequisites

Python 3.12+ and Node.js 20+
Neon PostgreSQL database (free tier works)
Google AI Studio API key (free)
Google OAuth client ID + secret

1 — Clone and set up backend

git clone https://github.com/your-username/munim.git
cd munim/backend

python3 -m venv venv
source venv/bin/activate        # Windows: venv\Scripts\activate
pip install -r requirements.txt

cp .env.example .env            # Then fill in your credentials

2 — Configure `.env`

# Database
DATABASE_URL=postgresql+asyncpg://user:pass@ep-xxx.neon.tech/neondb?sslmode=require

# Google AI (Gemini)
GOOGLE_API_KEY=AIza...

# Google OAuth
GOOGLE_CLIENT_ID=xxxx.apps.googleusercontent.com
GOOGLE_CLIENT_SECRET=GOCSPX-...
NEXTAUTH_SECRET=any-random-32-char-string

# WhatsApp via Twilio (sandbox — no business verification needed)
TWILIO_ACCOUNT_SID=ACxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
TWILIO_AUTH_TOKEN=your_auth_token
TWILIO_WHATSAPP_FROM=whatsapp:+14155238886

# App
APP_ENV=development
APP_SECRET_KEY=change-me-in-production

3 — Start backend

cd backend
python -m uvicorn main:app --reload --port 8000
# Swagger docs: http://localhost:8000/docs

4 — Set up and start frontend

cd frontend
npm install

# Create frontend/.env.local
NEXTAUTH_URL=http://localhost:3000
NEXTAUTH_SECRET=same-as-backend-secret
GOOGLE_CLIENT_ID=xxxx.apps.googleusercontent.com
GOOGLE_CLIENT_SECRET=GOCSPX-...
NEXT_PUBLIC_API_URL=http://localhost:8000/api/v1

npm run dev
# App: http://localhost:3000

WhatsApp Setup (5 minutes, free)

Option A — Twilio Sandbox (Recommended for testing)

Sign up at twilio.com — no credit card needed
Go to Messaging → Try it out → Send a WhatsApp message
Send the activation code to +1 415 523 8886 on WhatsApp
Copy Account SID and Auth Token from Console → Account Info
Paste into .env → done!

Option B — Meta Business API (Production)

Set WHATSAPP_PHONE_NUMBER_ID and WHATSAPP_ACCESS_TOKEN in .env. Munim automatically uses whichever provider is configured. Twilio takes priority if both are set.

Environment Variables

Variable	Required	Description
`DATABASE_URL`	✅	Neon PostgreSQL connection string
`GOOGLE_API_KEY`	✅	Gemini API key from Google AI Studio
`GOOGLE_CLIENT_ID`	✅	Google OAuth client ID
`GOOGLE_CLIENT_SECRET`	✅	Google OAuth client secret
`NEXTAUTH_SECRET`	✅	Random string for session signing
`TWILIO_ACCOUNT_SID`	⚡	Twilio Account SID (WhatsApp)
`TWILIO_AUTH_TOKEN`	⚡	Twilio Auth Token (WhatsApp)
`TWILIO_WHATSAPP_FROM`	⚡	Sandbox number — default: `whatsapp:+14155238886`
`WHATSAPP_PHONE_NUMBER_ID`	—	Meta Business API phone number ID
`WHATSAPP_ACCESS_TOKEN`	—	Meta Business API access token
`APP_ENV`	—	`development` or `production`
`APP_SECRET_KEY`	—	JWT signing secret

How AI Schema Detection Works

Munim sends only the column headers and 8 sample rows to Gemini — never the full dataset:

Headers:  ["Bill Date", "Invoice No", "Party Name", "Item Name", "Net Total"]
Samples:  [{"Bill Date": "2026-01-15", "Party Name": "Sharma Medical", ...}]

Gemini returns:
{
  "date_column":     "Bill Date",
  "amount_column":   "Net Total",
  "customer_column": "Party Name",
  "product_column":  "Item Name",
  "business_type":   "medical pharmacy",
  "confidence":      0.97
}

If confidence < 0.65 or Gemini is unavailable, falls back to rule-based pattern matching.

Project Structure

munim/
├── backend/
│   ├── routers/
│   │   ├── auth.py              # Google OAuth + profile management
│   │   ├── upload.py            # File upload + status polling
│   │   ├── analysis.py          # Metrics, anomalies, customers, history
│   │   ├── reports.py           # WhatsApp report generation + delivery
│   │   ├── ca.py                # CA firm portfolio endpoints
│   │   └── whatsapp.py          # Webhook + send endpoints
│   ├── services/
│   │   ├── ingestor/
│   │   │   ├── excel_parser.py           # Excel / CSV parsing
│   │   │   ├── tally_parser.py           # Tally XML parsing
│   │   │   ├── schema_detector.py        # Format detection + routing
│   │   │   └── gemini_schema_detector.py # AI column detection
│   │   ├── analytics/
│   │   │   ├── metrics.py       # Revenue, top products, dead stock
│   │   │   ├── anomaly.py       # Anomaly detection
│   │   │   ├── rfm.py           # RFM customer segmentation
│   │   │   ├── seasonality.py   # Seasonal context generation
│   │   │   └── ai_insights.py   # Gemini business insights
│   │   └── reporter/
│   │       └── llm_narrator.py  # WhatsApp report generation
│   ├── tasks/
│   │   └── process_upload.py    # Full analysis pipeline
│   ├── tests/                   # 122 pytest tests
│   └── config.py                # Settings via pydantic-settings
│
└── frontend/
    ├── app/
    │   ├── page.tsx                    # Public landing page
    │   ├── (auth)/login/               # Google sign-in
    │   ├── (dashboard)/
    │   │   ├── dashboard/              # Authenticated home with charts
    │   │   ├── upload/                 # Drag-and-drop file upload
    │   │   ├── analysis/[id]/          # Full business dashboard
    │   │   ├── reports/                # Upload history list
    │   │   ├── alerts/                 # Anomaly alerts
    │   │   └── settings/               # Profile + WhatsApp preferences
    │   └── (ca)/                       # CA firm portal
    ├── components/
    │   ├── layout/                     # AppShell, Sidebar, TopBar, MobileNav
    │   └── dashboard/                  # MetricCard, RevenueChart
    └── lib/
        ├── api/                        # React Query hooks
        └── types.ts                    # TypeScript interfaces

Running Tests

cd backend
source venv/bin/activate

pytest tests/ -q                       # Quick summary — 122 tests
pytest tests/ -v                       # Verbose with test names
pytest tests/test_metrics.py -v        # Revenue + analytics
pytest tests/test_anomaly.py -v        # Anomaly detection
pytest tests/test_tally_parser.py -v   # Tally XML parser

Contributing

Fork the repository
Create a feature branch: git checkout -b feat/your-feature
Make changes and run tests: pytest tests/ -q
Commit: git commit -m "feat: description"
Open a pull request

License

MIT — see LICENSE for details.

Built with ❤️ for Indian businesses

मुनीम — आपका भरोसेमंद डिजिटल हिसाबकिताब

Name		Name	Last commit message	Last commit date
Latest commit History 43 Commits
.github/workflows		.github/workflows
backend		backend
data/samples		data/samples
frontend		frontend
test_data		test_data
.env.example		.env.example
.gitignore		.gitignore
README.md		README.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

मुनीम · Munim

AI Business Intelligence for Indian Small Businesses

What It Does

Architecture

Features

🤖 AI-Powered Schema Detection

📊 Complete Business Analytics

✨ AI Business Insights

💬 WhatsApp Reports

🏪 Supported File Formats

Tech Stack

Quick Start

Prerequisites

1 — Clone and set up backend

2 — Configure `.env`

3 — Start backend

4 — Set up and start frontend

WhatsApp Setup (5 minutes, free)

Option A — Twilio Sandbox (Recommended for testing)

Option B — Meta Business API (Production)

Environment Variables

How AI Schema Detection Works

Project Structure

Running Tests

Contributing

License

About

Uh oh!

Releases

Packages

Uh oh!

Contributors

Uh oh!

Languages

Folders and files

Latest commit

History

Repository files navigation

मुनीम · Munim

AI Business Intelligence for Indian Small Businesses

What It Does

Architecture

Features

🤖 AI-Powered Schema Detection

📊 Complete Business Analytics

✨ AI Business Insights

💬 WhatsApp Reports

🏪 Supported File Formats

Tech Stack

Quick Start

Prerequisites

1 — Clone and set up backend

2 — Configure .env

3 — Start backend

4 — Set up and start frontend

WhatsApp Setup (5 minutes, free)

Option A — Twilio Sandbox (Recommended for testing)

Option B — Meta Business API (Production)

Environment Variables

How AI Schema Detection Works

Project Structure

Running Tests

Contributing

License

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

2 — Configure `.env`

Packages