A sandbox AI voice agent that simulates fraud alert calls for a bank. Designed for demo and learning purposes using fake data only.

This repository provides a step-by-step implementation of a fraud alert voice agent:

• 🤖 AI-powered fraud voice agent
• 🧾 Fake database of transactions
• 🎤 Text-to-speech (TTS) using Murf Falcon
• ✅ Verification flow (non-sensitive info)
• 📊 Transaction status logging
• ☎️ Optional LiveKit Telephony integration

Perfect for developers, AI enthusiasts, and anyone learning voice agent integration with banking-like workflows.

• JSON-based mock transactions
• Includes fake user name, card, amount, merchant, timestamp
• Simple security questions for verification

• Greets user as bank fraud representative

• Performs safe identity verification

• Reads suspicious transaction details

• Confirms legitimacy (yes/no)

• Updates transaction status in database:

  • confirmed_safe
  • confirmed_fraud
  • verification_failed

• Uses Murf Falcon API for natural AI voice
• Runs in browser without telephony
• Easy to extend for multiple cases

• Integrate with LiveKit Telephony
• Handle real phone calls for fraud verification
• Persist outcomes back to database

Watch the fraud alert agent in action:

🔗https://drive.google.com/file/d/1txCNPWyWqrocn4XY-J23j6d_56pdOzv1/view?usp=drive_link

Clone the repository:

git clone https://github.com/your-username/bank-fraud-alert-ai.git
cd bank-fraud-alert-ai

Install dependencies (Python example):

pip install -r requirements.txt

Run the voice agent (MVP):

python main.py

Follow the terminal prompts for username and verification flow.

/Bank-Fraud-Alert-AI
│
├── backend/
│   ├── fraud_cases.json
│   ├── main.py
│   └── utils.py
├── frontend/
│   ├── index.html
│   └── voice_agent.js
├── demo_videos/
├── README.md
└── requirements.txt

• Multiple fraud cases & advanced workflow
• DTMF input support for telephony
• AI-driven fraud scoring and alerts
• Real-time dashboard for case tracking
• Integration with SMS/email notifications

Om Gedam GitHub: @itsomg134 Email: omgedam123098@gmail.com Twitter (X): @omgedam LinkedIn: Om Gedam Portfolio: ogworks.lovable.app

Made with ❤️ for learners building AI-powered voice applications and fraud detection demos.

Do you want me to do that?