Overview
Today’s Agenda
Join us as we explore how AI is revolutionizing workplace seat booking through intelligent conversation and automation.
-
Project Overview
Exploring the background, challenges, and core objectives behind our AI powered solution -
Live Demonstration
Showcasing the AI chatbot in action within the Associate Arena app. -
Architecture Deep
Dive A comprehensive walkthrough of technical components and data flow -
Performance & Adoption
Analyzing user statistics and key performance indicators that demonstrate success -
Challenges & Future
Roadmap Discussing hurdles overcome and our ambitious vision for continuous improvement.
The Solution
The Vision: Simplifying Seat Booking
The deployment of the Project pipeline demonstrated
significant leaps in both classification accuracy and
strategic defect management.
The Challenge
Traditional seat booking systems create
unnecessary friction in the daily workflow.
Associates consistently face:
- Multiple clicks and complex navigation through outdated UIs
- Difficulty booking seats while mobile or on the go
- Valuable time wasted on repetitive administrative tasks
- Limited flexibility for multi-day or last minute bookings
Our AI-Powered Solution
An intelligent, voice-based chatbot
seamlessly integrated into the Associate
Arena app that enables users to book, view,
and manage office seats using natural
language.
Key Benefits
- Enhance User Experience: Frictionless, conversational interface
- Increase Efficiency: Reduce booking time by up to 80%
- Promote Flexibility: Enable easy multi-day and on-demand bookings
The Challenge
System Architecture Deep Dive
Our solution leverages cutting-edge AI and cloud technologies to deliver a robust, scalable booking experience
User Interaction
Associate makes a natural language request in the Associate Arena app: “Book my seat for the next three days”
Authentication Layer
App sends encrypted Auth Token to our FastAPI Endpoint for secure identity validation.
AI Core - LangGraph Agent
User message processed by LangGraph Agent powered by Azure OpenAI GPT-4o to understand intent and context
Intelligent Tool Selection
Agent selects the appropriate tool, for example, book_seat_for_multiple_dates, based on the analyzed intent.
API Execution
Selected tool calls the relevant backend API, for example, api/flexi/book4Days, to execute the booking action
Database Transaction
Azure SQL Database processes the request, confirms availability, and records the booking.
Confirmation Response
Success message returned to the user with booking details and confirmation number.
The Solution
The User Journey in a Nutshell
From voice command to confirmed booking experience the seamless flow that makes seat reservation effortless.
Strong Start: Analyzing User Adoption
Our AI-powered booking system has quickly gained traction, demonstrating clear value to
associates across all office locations.
Key Benefits
The tool demonstrates consistent daily engagement, with users returning regularly a strong
indicator of satisfaction and system reliability. Peak usage patterns align with typical weekly
planning cycles, validating our understanding of user behavior
0
Total Seats Booked
Fantastic initial adoption demonstrating strong user confidence in the AI system
0
ADU Office Bookings
Highest demand location showing
exceptional engagement with flexible
seating
0
Peak Daily Bookings
Recorded on September 1st as
associates planned their weekly
schedules
Learning and Evolving: Analyzing User Adoption
Issues Faced & Lessons Learned
- LLM Hallucination: Early versions occasionally generated incorrect information. Resolved by improving prompting techniques and combining redundant tools into streamlined functions.
- Voice Recognition Accuracy: Native Android and iOS features proved inconsistent. Successfully migrated to OpenAI Whisper model for superior voice-to-text conversion
- Initial Latency: Early requests experienced delays due to Less optimized backend system.
- Handling Ambiguity: Vague requests like “book me a seat next week” required sophisticated clarification logic.
- API Error Propagation: Backend failures resulted in unclear error messages being shown to users.
Improvements & Future Roadmap
- Performance Tuning: Implemented intelligent caching for common queries, reducing average latency by 65%.
- Robust Error Handling: Developed user-friendly error messages with automatic retry mechanisms
Exciting Future Features:
- QR codes and file upload in the chatbot to book a seat.
- Proactive booking notifications and reminders
- Meeting room bookings with calendar integration
- User feedback system (right/worng) for continuous improvement
