Agentix
UA Innovate Hackathon • Winner
The Problem
Customer support teams were overwhelmed with repetitive ticket management tasks - manually assigning tickets to appropriate agents, sending status notifications, and closing resolved issues. This manual process led to delayed responses and inconsistent customer service quality.
Challenge
Design and build an intelligent ticketing system in a hackathon timeframe that could automate the entire ticket lifecycle while maintaining accuracy and providing real-time updates to both support teams and customers.
Solution
We developed an intelligent support system that completely automates ticket management using AI. The system analyzes incoming tickets (including images), automatically classifies and assigns them to appropriate agents, sends real-time notifications via WebSockets, and can even auto-resolve simple issues, all while keeping customers informed through automated email updates.
Process
- Problem Analysis: Analyzed existing support ticket workflows and identified automation opportunities. Researched AI tool-calling capabilities and real-time notification requirements.
- Core AI Development: Built Gemini AI integration for automatic ticket classification and agent assignment. Implemented image analysis for visual support tickets.
- Real-time System: Integrated Firebase for live updates and WebSocket connections for instant notifications. Created automated email workflows for ticket lifecycle events.
- Testing & Demo: End-to-end testing of automated workflows and real-time updates. Prepared comprehensive demo showing complete ticket lifecycle automation.
Challenges Overcome
- AI Accuracy for Ticket Classification: Fine-tuned Gemini prompts and implemented confidence scoring to ensure accurate ticket routing and escalation when needed
- Real-time Synchronization: Combined Firebase real-time database with WebSocket connections to ensure all stakeholders receive instant updates
- Image-based Ticket Processing: Integrated Gemini's vision capabilities to analyze screenshots and visual issues, enabling automated technical support
Impact & Results
- Ticket Processing: Manual assignment/updates → Fully automated workflow (End-to-end automation)
- Response Time: Manual notification delays → Instant real-time updates (Real-time communication)
- Support Capabilities: Text-only tickets → Image-guided analysis (Enhanced support options)
Key Achievements
- Won UA Innovate hackathon with innovative AI-driven support automation
- Built comprehensive ticket lifecycle automation from assignment to closure
- Implemented real-time updates using Firebase and WebSocket notifications
- Created image-guided support system using Gemini AI for visual issue analysis
Tech Stack
Firebase, WebSockets, Google Gemini AI, Tool Calling API, Image Analysis, Real-time Database, Automated Email Systems, AI Classification