An autonomous voice agent that scrapes real estate leads, matches them with potential buyers, makes calls, closes deals, and makes money while you sleep.

This tool creates an AI-powered sales representative that:

1. Scrapes real estate agent leads from platforms like Realtor.com and Zillow
2. Mines buyer intent signals from Twitter, Reddit, and Facebook groups
3. Matches buyers to agents based on location and specialization
4. Makes real phone calls using natural-sounding AI voice
5. Handles live conversations with GPT-4 intelligence
6. Closes sales autonomously and logs everything to your dashboard
7. Runs 24/7 - works while you sleep, doesn't take breaks, never complains

In a real test with real estate agents:

• 38 outbound calls
• 11 live conversations
• 4 paid signups at $149 each
• $596 in revenue
• $578 profit after costs ($18 total for API calls)
• 10.5% conversion rate (3x industry average)
• TCPA compliant with clear disclosures

git clone https://github.com/llSourcell/my_ai_intern.git
cd my_ai_intern

Create a .env file in the /backend directory:

TWILIO_ACCOUNT_SID=your_twilio_sid
TWILIO_AUTH_TOKEN=your_twilio_auth_token
TWILIO_PHONE_NUMBER=your_twilio_phone_number
ELEVENLABS_API_KEY=your_elevenlabs_api_key
ELEVENLABS_AGENT_ID=h3dC4sQ9cPDtYItAe0Z8
LLM_API_KEY=your_openai_api_key
BRIGHTDATA_API_TOKEN=your_brightdata_api_token
BRIGHTDATA_WEB_UNLOCKER_ZONE=your_zone_name
DB_TYPE=sqlite
DATABASE_URL=sqlite:///leads.db

Where to get API keys:

• Twilio: Sign up at twilio.com (provides phone capabilities)
• ElevenLabs: Sign up at elevenlabs.io (provides realistic voices)
• OpenAI: Sign up at openai.com (powers the AI conversations)
• Bright Data: Sign up at brightdata.com - receive $10 free MCP credits (powers web scraping)

# Install backend dependencies
cd backend
pip install -r requirements.txt

# Install frontend dependencies
cd ../frontend
npm install

# Start the backend (from the backend directory)
python app.py

# In a new terminal, start the frontend (from the frontend directory)
npm start

The dashboard will open at http://localhost:3000

1. Configure your API keys in the dashboard settings
2. Click "Scrape New Leads" to gather real estate agents
3. The system will automatically match agents with potential buyers from social media
4. Select leads and click "Call" to initiate automated outreach
5. Monitor conversations and results in real-time
6. View call logs and transcripts for each lead

1. Scrape Real Estate Agents: Uses Bright Data's MCP to pull agent data from Realtor.com and Zillow

   # MCP servers provide stealth mode for bots - unblockable real-time web access
   # Using Bright Data's Web Unlocker to access agent listings
   agents = scrape_agents_via_mcp("Austin, TX", limit=100)

2. Mine Buyer Intent: Searches social media for people looking to buy/sell

   # Search for posts like "Moving to Austin" or "Need a 3-bed in 78704"
   buyer_signals = mine_intent_signals(["twitter", "reddit", "facebook_groups"])

3. Match Buyers to Agents: Connects buyers with agents in the same areas

   # Match based on zip code, property type, and specialization
   matched_leads = match_buyer_to_agent(buyer_signals, agents)

Modify the prompt in voice.py to change the sales approach:

# Customize your sales script here
script = f"Hi {agent_name}, this is Ava from Home IQ. We've tracked {buyer_count} qualified buyers searching in {zip_code} right now. Would you like a list?"

• Backend: Flask, SQLite, Bright Data MCP for scraping
• APIs: Twilio (calls), ElevenLabs (voice), OpenAI (GPT-4 intelligence)
• Frontend: React with Tailwind UI
• Performance: Automated scheduling, call logging, analytics

This project uses Bright Data's Model Context Protocol (MCP) servers for powerful web scraping:

• Unblockable Access: MCP servers provide stealth mode that can bypass CAPTCHAs and blocks
• Global Coverage: Access data from anywhere with worldwide IP coverage
• AI-Ready Data: Automatically formats extracted content for AI processing
• Dynamic Content: Handles JavaScript-rendered content effortlessly

// Example MCP Server connection
const SBR_CDP = 'wss://brd-customer-CUSTOMER_ID-zone-ZONE_NAME:[email protected]:9222';

async function scrapeAgents() {
    const browser = await pw.chromium.connectOverCDP(SBR_CDP);
    const page = await browser.newPage();
    await page.goto('https://www.realtor.com/realestateandhomes-search/Austin_TX/pg-1');
    // Extract agent data...
}

Once your MVP is working, consider these scaling approaches:

1. Vertical expansion: Target different industries (e-commerce, insurance, etc.)
2. Recurring revenue: Convert one-time sales ($149) to subscriptions ($99/month)
3. Self-optimization: Let the AI improve pitches based on call outcomes

This tool is designed with TCPA compliance in mind:

• Each call opens with a clear disclosure
• Provides opt-out options
• Logs all interactions

Always verify regulations in your region before automated calling.

• Check the code examples in each file
• Consult the API documentation for Twilio, ElevenLabs, OpenAI, and Bright Data
• Open an issue on GitHub

Built with ❤️ using Flask, React, and AI APIs