OpenClaw India Guide: WhatsApp, Local Models & INR Costs

OpenClaw fits India better than any other AI personal assistant on the market. It is free and open-source at a time when most AI tools charge $20/month in USD. It works natively with WhatsApp — the app through which India communicates. It runs on local hardware, keeping your data in India. And the community has built dozens of skills specifically for Indian workflows: IRCTC train tracking, UPI reminders, GST invoicing, BSE/NSE alerts, and more.

This guide covers everything you need to get OpenClaw running in India, from zero to a fully automated personal assistant, with a focus on keeping costs as low as possible.

Why OpenClaw Is Built for India

WhatsApp integration: India has over 500 million WhatsApp users. No other AI assistant connects directly to WhatsApp — you can message OpenClaw from your existing WhatsApp number and it responds and acts in that same conversation.

Zero subscription cost: Most AI assistants charge $10–$30/month. That is ₹840–₹2,500/month just for one AI tool. OpenClaw is MIT-licensed open-source software. You pay nothing for the software. You can also run it at ₹0/month total using free models.

Privacy and data sovereignty: When you run OpenClaw with Ollama locally, your data never leaves your device or India. No American or European company processes your messages, task history, or documents.

Works on any hardware: You do not need a powerful machine. OpenClaw itself is lightweight. Even older laptops (2018+, 8 GB RAM) can run it effectively with Groq's free cloud API.

India-specific skills ecosystem: ClawHub's India category has 310+ skills for IRCTC, UPI, Indian stock markets, GST compliance, cricket, and Indian news.

How to Set Up OpenClaw for Free in India

Step 1: Install Node.js 22

# Ubuntu / WSL2 (most common for India users)
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt-get install -y nodejs

# macOS (for MacBook users)
brew install node@22

# Verify
node --version  # v22.x.x

Step 2: Install Ollama (Free Local AI Runtime)

Ollama lets you run AI models locally — no internet needed for inference, no API costs:

# Linux / WSL2
curl -fsSL https://ollama.com/install.sh | sh

# macOS
brew install ollama

Start the Ollama server:

ollama serve

Leave this running in a terminal tab. Ollama listens at http://localhost:11434.

Step 3: Download Llama 3.2 (Completely Free, Runs on 8 GB RAM)

# Llama 3.2 8B — good balance of quality and RAM usage
ollama pull llama3.2

# Or Llama 3.3 70B for better quality (requires 16 GB RAM)
ollama pull llama3.3

# Test it works
ollama run llama3.2 "Namaste! Aap kya kar sakte ho?"

Llama 3.2 downloads are 4–5 GB. On Indian broadband, this takes 15–30 minutes. It is a one-time download.

Step 4: Install OpenClaw and Configure Ollama as AI Provider

npm install -g openclaw clawhub
mkdir ~/my-assistant && cd ~/my-assistant
openclaw init

When the init wizard asks for your AI provider, choose "Ollama". Then verify openclaw.config.json looks like this:

{
  "ai": {
    "provider": "ollama",
    "model": "llama3.2",
    "baseUrl": "http://localhost:11434"
  },
  "memory": {
    "enabled": true,
    "path": "~/clawd/memory"
  },
  "channels": []
}

No API key needed for Ollama.

Step 5: Set Up a Telegram Bot (Simplest Channel for India)

Telegram is the easiest starting channel — no business account, no approval, works in under 5 minutes:

1. Open Telegram, search @BotFather
2. Send /newbot
3. Choose a name (e.g., "My Assistant") and username (e.g., myassistant_bot)
4. BotFather gives you a token: 7123456789:AAH...
5. Add to your .env file:

TELEGRAM_BOT_TOKEN=7123456789:AAH...your_token_here

6. Update openclaw.config.json:

{
  "channels": [
    {
      "type": "telegram",
      "tokenEnv": "TELEGRAM_BOT_TOKEN",
      "allowedUsers": ["your_telegram_username"]
    }
  ]
}

Step 6: Test with a Hindi Message

Start OpenClaw:

openclaw start

Go to your Telegram bot and send:

Namaste! Kya tum mujhe help kar sakte ho?

You should get a Hindi response. Your agent is live.

---

WhatsApp Business API Setup for India

WhatsApp integration requires more steps but is worth it for most Indian users — it connects your AI agent to the app everyone already uses.

What you need:

• Meta Business account (free)
• A phone number dedicated to your AI assistant (a spare SIM works)
• A server with a public URL for webhooks (or ngrok for testing)

Setup steps:

1. Go to developers.facebook.com and create a developer account
2. Create a new app, choose "Business" type
3. Add the "WhatsApp" product to your app
4. In WhatsApp > Getting Started, add your phone number
5. Complete business verification (takes 1–3 business days in India)
6. Generate a permanent access token from System Users in Meta Business Settings
7. Add your credentials to .env:

WHATSAPP_ACCESS_TOKEN=your_permanent_token
WHATSAPP_PHONE_NUMBER_ID=your_phone_number_id
WHATSAPP_VERIFY_TOKEN=your_custom_verify_string

8. Configure the webhook: In Meta Developer Console, set the webhook URL to https://your-server.com/webhook/whatsapp and the verify token to match WHATSAPP_VERIFY_TOKEN

9. Add WhatsApp to openclaw.config.json:

{
  "channels": [
    {
      "type": "whatsapp",
      "accessTokenEnv": "WHATSAPP_ACCESS_TOKEN",
      "phoneNumberIdEnv": "WHATSAPP_PHONE_NUMBER_ID",
      "verifyTokenEnv": "WHATSAPP_VERIFY_TOKEN"
    }
  ]
}

Meta's free tier in India: 1,000 user-initiated conversations per month at ₹0. After that, pricing depends on conversation type (₹0.35–₹0.70 per conversation approximately, varying by month). For personal use, you will almost certainly stay under the free tier.

For testing: Use ngrok to expose your local OpenClaw to the internet temporarily:

ngrok http 18789
# Use the https URL ngrok gives you as your webhook URL

---

INR Cost Comparison: All AI Model Options

| Model | Provider | Cost per 1K tokens | Monthly cost (50 queries/day) | |---|---|---|---| | Llama 3.2 via Ollama | Local | ₹0 | ₹0 | | Llama 3.3 via Ollama | Local | ₹0 | ₹0 | | Llama 3.1 70B via Groq | Cloud (free tier) | ₹0 | ₹0 | | Gemini 1.5 Flash | Google Cloud | ~₹0.25 | ~₹100–300 | | DeepSeek V3 | DeepSeek API | ~₹0.27 | ~₹110–330 | | GPT-4o mini | OpenAI | ~₹1.25 | ~₹500–1,500 | | GPT-4o | OpenAI | ~₹8 | ~₹3,200–9,600 | | Claude 3 Haiku | Anthropic | ~₹0.63 | ~₹252–756 | | Claude Sonnet 4 | Anthropic | ~₹4 | ~₹1,600–4,800 |

Note: Costs are approximate and vary with context length. Token counts per query vary significantly based on conversation length and skill complexity.

Recommended approach for India:

• Start with Groq free tier (zero cost, cloud speed)
• Add Ollama as fallback for when Groq rate limits hit
• Only add paid models if you specifically need their capabilities

---

Hindi Language Configuration

OpenClaw's language is determined by your system prompt and the AI model you use. Configure a Hindi-capable system prompt:

In openclaw.config.json, add a systemPrompt section:

{
  "systemPrompt": "Aap ek personal AI assistant hain. Jab bhi user Hindi mein baat kare, Hindi mein jawab dein. Jab English mein baat kare, English mein jawab dein. Hinglish bhi theek hai. Aap India mein hain, isliye IST timezone use karein, currency INR (₹) mein batayein, aur Indian context samjhein."
}

Or for a pure English assistant that understands Hindi inputs:

{
  "systemPrompt": "You are a personal AI assistant for an Indian user. Always respond in English, but understand and process Hindi, Hinglish, and regional language inputs. Use IST timezone, INR for currency, and Indian context throughout."
}

Hinglish prompts work naturally. You can send messages like:

• "Kal ka weather kya hai Bangalore mein?"
• "IRCTC ka PNR 4512345678 check karo"
• "Mujhe Friday 3 PM ko meeting ka reminder set karo"
• "Email draft karo HR ko regarding medical leave"

All current major models (Llama 3.2, Gemini, Claude) handle Hinglish very well.

---

India-Specific Automations

Train Status Monitoring (IRCTC)

Install the IRCTC skill:

clawhub install irctc-train-alerts

Usage:

Check PNR status for 4512345678 and notify me if it changes
Monitor train 12951 Mumbai Rajdhani — alert me if it's delayed by more than 30 minutes
Set up daily departure reminders for my weekly Mumbai-Pune commute on train 12127

The skill fetches data from IRCTC's status API and sends you Telegram/WhatsApp messages when status changes.

UPI Payment Reminders and Tracking

clawhub install upi-reminder

Set up recurring reminders:

Remind me every month on the 5th to pay credit card bill
Remind me on the 15th to pay electricity bill of approximately ₹1,200
Track that I owe Suresh ₹3,500 for the birthday dinner — remind me on Friday

BSE/NSE Stock Price Alerts

clawhub install bse-nse-alerts

Usage:

Alert me if Infosys drops below ₹1,400 or rises above ₹1,600
Send me a daily 9:30 AM BSE opening summary for my watchlist: RELIANCE, TCS, INFY, HDFC
Alert me if any stock in my watchlist moves more than 5% in a session

Cricket Match Alerts

clawhub install cricket-alerts

Usage:

Send me live score updates for India vs Australia Test starting tomorrow
Alert me when India's wickets fall during any ongoing match
Send me the IPL schedule for this week

GST Filing Deadline Reminders

clawhub install gst-invoice-helper

OpenClaw automatically tracks GST filing deadlines based on your configured GSTIN and sends you reminders:

Add my GSTIN 27AAPFU0939F1ZV and track all filing deadlines
Remind me 3 days before every GSTR-1 and GSTR-3B due date
Generate a GST invoice for ₹85,000 + 18% GST for web development services to Mehta Enterprises, Mumbai

Morning India Briefing

clawhub install india-news-briefing

Configure your morning routine:

Every day at 7 AM IST, send me:
1. Today's weather in Bengaluru
2. Top 5 Indian tech and startup news from Economic Times and NDTV
3. BSE Sensex and Nifty 50 opening levels
4. My first 3 calendar events for today
5. Any pending UPI reminders due today
Keep it under 300 words. Send to Telegram.

---

Best Free Models for India: Comparison

| Model | Where to Run | RAM Needed | Quality | Best For | |---|---|---|---|---| | Llama 3.2 8B | Ollama (local) | 8 GB | Good | Daily tasks, reminders, Hindi | | Llama 3.3 70B | Ollama (local) | 16 GB | Excellent | Complex tasks, analysis | | Llama 3.1 70B | Groq (free cloud) | None | Excellent | Best free cloud option | | Gemini 1.5 Flash | Google API | None | Very Good | Low-cost cloud, multimodal | | DeepSeek V3 | DeepSeek API | None | Excellent | Cheapest capable cloud model |

Recommendation for 8 GB RAM machines: Groq free tier as primary, Llama 3.2 via Ollama as offline fallback. You get cloud-quality AI most of the time at zero cost.

Recommendation for 16 GB RAM machines: Llama 3.3 70B via Ollama as primary. Near GPT-4 quality, completely free, fully private. Use Groq free tier when you need faster responses.

Recommendation for business use with sensitive data: Ollama exclusively. Data never leaves India. DPDP Act 2023 compliant.

---

Running OpenClaw 24/7 from India

Your laptop will not always be on. For 24/7 automation (morning briefings, background monitoring, price alerts), you need a server:

Cheapest options in India:

• DigitalOcean Mumbai (BLR1 region): $6/month (~₹500/month) for a 1 GB RAM droplet — enough for OpenClaw with Groq API
• AWS EC2 t4g.micro, ap-south-1 Mumbai: Free for 12 months under AWS Free Tier, then ~₹400/month
• Hetzner Helsinki/Nuremberg (international but cheap): €3.79/month (~₹330/month) for a 2 GB RAM VPS — great value
• Home router/Raspberry Pi 4: One-time cost ₹4,500–₹6,000, then ₹0/month ongoing

On a cloud VPS, run OpenClaw as a systemd service so it starts automatically after reboots:

# /etc/systemd/system/openclaw.service
[Unit]
Description=OpenClaw AI Agent
After=network.target

[Service]
User=ubuntu
WorkingDirectory=/home/ubuntu/my-assistant
ExecStart=/usr/bin/openclaw start
Restart=on-failure
RestartSec=10
Environment=NODE_ENV=production

[Install]
WantedBy=multi-user.target

sudo systemctl enable openclaw
sudo systemctl start openclaw
sudo systemctl status openclaw

Your agent now runs 24/7, sending you morning briefings, monitoring prices, and processing automation tasks — even when your laptop is off.

---

A Complete India Power User Setup

Here is a config that combines everything above for a typical Indian professional:

{
  "ai": {
    "routing": {
      "default": {
        "provider": "groq",
        "model": "llama-3.1-70b-versatile",
        "apiKeyEnv": "GROQ_API_KEY"
      },
      "overflow": {
        "provider": "ollama",
        "model": "llama3.2",
        "baseUrl": "http://localhost:11434"
      }
    },
    "failover": ["groq", "ollama"]
  },
  "systemPrompt": "You are a personal AI assistant for an Indian professional. Use IST timezone. Show costs in INR. Understand Hinglish naturally. Be concise.",
  "memory": {
    "enabled": true,
    "provider": "pearl",
    "path": "~/clawd/memory"
  },
  "channels": [
    {
      "type": "telegram",
      "tokenEnv": "TELEGRAM_BOT_TOKEN",
      "allowedUsers": ["your_telegram_username"]
    }
  ],
  "crons": [
    {
      "name": "morning-india-briefing",
      "cron": "0 7 * * 1-6",
      "task": "Send morning briefing: weather Bengaluru, top 5 Indian tech news, Sensex/Nifty opening, my first 3 calendar events"
    },
    {
      "name": "market-close",
      "cron": "30 15 * * 1-5",
      "task": "Send BSE/NSE closing summary for my watchlist"
    }
  ]
}

This setup delivers: free AI routing with local fallback, Hindi/Hinglish understanding, morning briefings, market alerts, IRCTC and UPI automation, GST invoice generation — for an estimated ₹0–₹100/month depending on usage.

Where to Go Next

• OpenClaw Setup Guide — Full installation walkthrough for beginners
• OpenClaw Skills Guide — All India-specific skills explained
• OpenClaw Hacks and Tips — Advanced cost reduction and automation patterns
• OpenClaw Security Guide — DPDP Act compliance and data protection
• Compare OpenClaw vs Claude Cowork — Which is right for your use case?
• OpenClaw on PromptAndSkills — India-optimized prompts and skill templates