Global Guide Bot

• 

  Diagram

Inspiration

I witnessed tourists struggling to find authentic local experiences while talented guides lacked visibility. Traditional platforms are impersonal and expensive. When Amazon Nova Pro launched, I realized we could create an AI system that understands travel intent semantically through WhatsApp — the world's most accessible messaging platform.

What it does

Global Guide Bot transforms tourism through intelligent WhatsApp conversations:

• Text booking: "I need a temple guide in Bangkok" → AI semantic analysis → matched guides
  
• Image discovery: Upload travel photos → Nova Pro analyzes location/culture → personalized recommendations
  
• Cultural intelligence: Real-time etiquette guidance and cultural context
  
• Guide registration: Two-sided marketplace for local guides
  
• Memory & Strands: Multi-threaded conversations with persistent context

How I built it

Enhanced AgentCore Architecture with 5 specialized AI agents:

• Tourist Agent: Enhanced intent detection with booking confirmation logic
• Cultural Agent: Cultural intelligence using Nova Pro vision/text
  
• Guide Agent: Semantic matching with 200+ Thai guide database
  
• Booking Agent: Pricing breakdown and reservation management
  
• Registration Agent: Guide onboarding and validation
  

Tech Stack: AWS Lambda, DynamoDB, Nova Pro, API Gateway, Terraform IaC, WhatsApp Business API, Cloudfront

Challenges I ran into

• Agent Communication: Solved complex context sharing between agents using Memory & Strands architecture for conversation threading.
  
• AI Reliability: Balanced intelligence vs. hard corded keywords — AI for business logic, keywords for system control like yes/no/reset/clear/confirm.

Accomplishments that I'm proud of

• ✅ 5 Complete Demo Flows working end-to-end with real AWS integrations
  
• ✅ Production-Ready System with monitoring, error handling, and Terraform automation
  
• ✅ Memory & Strands Architecture enabling complex multi-threaded travel conversations
  

What I learned

• AI Integration: Semantic analysis dramatically improves user experience but requires fallback mechanisms for reliability.
  
• Conversation Architecture: Persistent memory and intelligent threading are crucial for complex travel planning scenarios.
  
• Agent Orchestration: Explicit context passing and semantic intent detection enable smooth agent handoffs.
  
• Production Deployment: Terraform automation and comprehensive monitoring are essential for scalable AI systems.

What's next for Global Guide Bot

Immediate: Multi-language support, real-time guide tracking, payment integration
Scaling: Global destination expansion, B2B hotel partnerships, tourism analytics
Technical: Voice message support, predictive recommendations, calendar integration Integration: Other chat platform e.g. LINE and also calendar platform Mobile Application: It may worth to have it own mobile application option with built it messaging system.

Built With

• amazonbedrock
• amazonqdeveloper
• apigateway
• cloudfront
• dynamodb
• kiro
• lambda
• python
• s3