Tulasi: Conversational Agent for Railway Enquiry
A conversational AI agent designed to help users get railway information quickly and easily through natural language interaction.
2020 • Conversational AI • UX Design
Context & Challenge
The Problem
Railway enquiry systems are notoriously complex and difficult to navigate. Users struggle with:
- Complex menu structures requiring multiple steps to find information
- Technical terminology that confuses non-technical users
- Limited availability of customer service representatives
- Fragmented information across different platforms
- Language barriers for users who prefer regional languages
Design Challenge
Create a conversational interface that understands user queries in natural language and provides accurate, timely information about train schedules, availability, and other railway services. The solution needed to be accessible to users of all technical backgrounds and support multiple interaction patterns.
Role & Tools
My Role
As the UX Designer, I was responsible for:
- User research and persona development
- Conversational flow design and dialogue mapping
- Natural language interface design
- Prototyping and user testing
- Collaboration with NLP engineers
Tools & Methods
- Figma - Interface design and prototyping
- Miro - Conversation flow mapping
- UserTesting - Usability testing
- Natural Language Processing - Intent recognition and entity extraction
Approach & Solution
Conversational Workflow Design
We designed a natural language interface that allows users to ask questions in plain language, similar to how they would ask a railway employee:
User asks:
"What trains go from Mumbai to Delhi tomorrow?"
Tulasi responds:
Lists available trains with times, duration, and availability, with options to book or get more details
Key Design Features
Natural Language Understanding
The system understands various phrasings of the same question and can handle follow-up queries with context awareness.
Multi-turn Conversations
Users can refine queries through conversation, asking follow-up questions without repeating context.
Visual + Text Responses
Information is presented both conversationally and visually, with train schedules, maps, and booking options.
Error Handling
When the system doesn't understand, it asks clarifying questions rather than showing errors.
Dashboard Wireframe
The interface combines conversational elements with traditional dashboard components for users who prefer visual navigation:
AI Automation Flow
The system uses AI to automate common queries and provide intelligent responses:
Impact & Results
User Experience Improvements
- Reduced query time from 3-5 minutes to under 30 seconds
- 90% of users successfully completed queries on first attempt
- High satisfaction with natural language interaction
- Reduced need for customer service support
UX Research Insights
User testing revealed key insights about conversational design for railway services:
Conversational Design Principles
This project established key principles for designing conversational interfaces:
Reflection
Tulasi demonstrated how conversational design can make complex information systems accessible to all users. By allowing natural language interaction, we removed the barrier of learning complex menu structures and technical terminology. This project laid the foundation for my later work in AI-driven SaaS design and conversational interfaces.
📸 Image Assets
To complete this case study, please add the following images to public/images/tulasi/:
- 1.png - Full conversational workflow diagram (Page 10 from PDF)
- 2.png - SaaS dashboard wireframe section (Page 15 from PDF)
- 3.png - AI automation flow diagram (Page 18 from PDF)
- 4.png - UX research insights infographic (Page 22 from PDF)
- 5.png - Conversational design principles diagram (Page 25 from PDF)
Images will automatically appear once placed in the directory. Use high-resolution PNG or JPEG format optimized for web.