🏆 Second Prize – AgentX AI Support Automation Challenge 2025
BusBuzz is an award-winning AI agent built for automated intercity bus ticket support.
Designed with advanced prompt engineering, controlled reasoning paths, and realistic customer interaction flows, BusBuzz secured Second Prize🥈 in the AgentX Challenge 2025.
BusBuzz is designed for intercity bus operators or ticketing platforms that need a customer support agent to handle booking questions, discount logic, seat availability concerns, and complaint escalation in a safe and realistic way.
- Multi-step reasoning for complex ticket queries
- Human-like conversation design tailored for Bangladeshi travel contexts
- Rush-hour seat scarcity logic without manipulative wording
- Clean discount system with rounded final prices
- Complaint escalation workflows
- Privacy-safe responses and controlled output boundaries
- Custom-built Prompt Architecture
- Defined system behaviors and linguistic constraints
- Optimized for fast response, controlled creativity, and high accuracy
- Modular design for future integrations (API + UI)
- Python
- OpenAI / LLM APIs
- Prompt Engineering
- Structured conversation design
- Modular workflow architecture
BusBuzz ranked Second Place at the AgentX AI Support Automation Challenge, recognized for:
- Consistent reasoning
- Natural conversational tone
- Clear constraint handling
- Robust discount and escalation logic
/prompts— core prompt system and logic/flows— ticketing, discount, and complaint scenarios/docs— design notes and behaviour constraints/examples— sample conversations and expected outputs
This project demonstrates expertise in crafting controlled, production-ready AI agents with:
- Prompt engineering
- Business logic design
- Customer support workflows
- Safety and constraint handling
- Live API integration
- User interface (web/app)
- Multi-language support (Bangla + English)
Python Data Cleaning Automation
Automation tool for cleaning messy Excel/CSV datasets using Python.
Repository: https://github.com/sayeduzzaman-bracu/excel-data-cleaning-automation