🚀 About the Project – FraudGuardAI 🛡️

🎯 Inspiration

In today’s digital banking era, financial fraud is on the rise — and traditional customer support systems often fail to respond quickly or efficiently.
This inspired me to build FraudGuardAI — a smart, responsive, and secure AI-powered fraud support assistant using Salesforce's Embedded Messaging.

I wanted to create a chatbot that helps users:

• Report unauthorized transactions 🏦
  
• Flag suspicious emails or account access issues ⚠️
  
• Seamlessly connect to a fraud specialist when needed 🤝

🧠 What I Learned

This was my first-ever experience building a Salesforce-based chatbot, and I learned a lot:

• How to use Salesforce Embedded Messaging from scratch
  
• How to design dynamic dialogs and context variables inside the bot
  
• How to configure Omni-Channel routing and Embedded Service Deployments
  
• How to debug CORS issues, live test bots on localhost, and use browser DevTools effectively
  

🛠️ How I Built It

• Set up a Salesforce Messaging Channel and Embedded Deployment.
• Created dialogs for 3 key fraud scenarios:
  • Unauthorized Transactions
    
  • Account Access Issues
    
  • Suspicious Emails or Messages
    
• Added Transfer to Agent logic with a simulated typing delay.
• Used Visual Studio Code and the Live Server extension to simulate local hosting.
• Embedded the bot in a local HTML test page using Salesforce’s JavaScript bootstrap library.
• Configured custom entities, context variables, and bot rules for a smooth conversation flow.

🧩 Challenges I Faced

• Getting the chat icon to appear on localhost required allowlisting domains and publishing correctly.
• Faced CORS issues and ERR_FAILED errors due to configuration mistakes.
• Creating a custom "Yes/No" entity using Regex was tricky and required precise formatting.
• Dialog logic had to be carefully structured to support both automation and agent handoff.

✅ Final Result

FraudGuardAI is now a working AI fraud assistant that:

• Helps users report fraud or suspicious activity
• Guides them with smart dialogs and questions
• Escalates to a live fraud agent when needed
• Runs securely on Salesforce with real-world use-case logic

🤖 This was an incredibly rewarding build — pushing me to explore Salesforce tools, understand enterprise-level messaging, and think like a fintech solution engineer.
I’m proud of the result — and excited for what's next!

Built With

• browserdevtools
• customcontextvariablesandentities
• html5
• javascript
• liveserver
• markdown
• regexbasedentitymatching
• salesforcebootstrapscript
• salesforcetools
• vscode