Designing an AI-driven travel chatbot
for smarter itinerary planning
PRODUCT DESIGN . CONVERSATIONAL UI
Won the People’s Choice Award at IUI Capstone Showcase
CONTEXT
This project was done in collaboration with LocHist LLC, a company focused on enhancing travel experiences through location-based storytelling. We designed an app that creates personalized itineraries with historical audio stories.
MY ROLE
I led the research, prototype, and usability testing of the AI Chatbot for itinerary planning.
TEAM
Five HCI graduate students, Indiana University Indianapolis
COMPANY
LocHistt LLC
Understanding the project as a whole
PROBLEM AREA
Traveling is fun, but planning a trip can be overwhelming and time consuming. Travelers often relying on multiple apps to gather scattered information while planning a trip.
85% of travelers
prefer a personalized experience aligned with their interests.
75% of travelers
want to learn about the destination in a fun and engaging way.
SOLUTION OVERVIEW
EnRoute is a smart travel companion that curates personalized itineraries based on user preferences, making trip planning effortless and engaging. By blending historical insights with immersive storytelling, EnRoute transforms the way travelers experience a destination, through tailored recommendations and interactive experiences.
Breaking down the core features of the application–
Breaking down the core features of the application–
01
Personalized Onboarding
Personalized Onboarding
Helps users choose their interests and needs, and to provide personalized recommendations.
Provides users with location-based audio stories, bringing history and culture to life as they explore.
03
AI driven Itinerary planner
AI driven Itinerary planner
Creates personalized itineraries using AI, tailoring destinations and activities to user preferences.
HIGH LEVEL BUSINESS GOALS
To leverage conversational AI to simplify and personalize trip planning
To include AI personalization for users to easily discover destinations
To gather user preferences and generate customized itineraries
DESIGN PROCESS
Since the project was divided into three core features, the design process followed a non-linear, iterative approach.
The AI chatbot design focused heavily on research, dialogue testing, and usability feedback. Throughout the process, I collaborated closely with stakeholder, incorporating feedback to make sure the chatbot balanced user control with smart suggestions.
The AI chatbot design focused heavily on research, dialogue testing, and usability feedback. Throughout the process, I collaborated closely with stakeholder, incorporating feedback to make sure the chatbot balanced user control with smart suggestions.
Research & Learnings That Shaped the Itinerary Planner
Before diving into the design, I first deepened my understanding of Conversational AI, frameworks, and best practices. I also interviewed potential travelers to understand their needs and frustrations.
KEY INSIGHTS
Conversational frameworks & guidelines
I gained a deeper understanding of chatbot capabilities, limitations, and how to leverage platform features for a better user experience. I explored Conversational Design and Natural Language Understanding (NLU), learning how intent recognition, entity extraction, and sentiment analysis create a more human-like chatbot.
Analyzing user pain points
I interviewed 10 participants to understand user behaviors and pain points in itinerary planning. I explored platforms like TripIt and Kayak, and recognized they lack flexibility for quick edits and personalization, while tools like Roadtrippers focus on logistics but lack interactive, conversational experiences.
SOLUTION
Designing the chatbot simulation
While the design process focused on getting the chatbot flow right, it was equally user-centric. I constantly referred back to user pain points at each stage, shaping a scenario-based solution that directly addresses the frustrations users face with existing itinerary planning tools.
USER PAIN POINT
"I want the chatbot to feel natural, but it sounds robotic and scripted."
SOLUTION
The chatbot speaks in a friendly, human-like tone, making the experience feel like chatting with a helpful guide instead of a machine.
This applies Grice’s Maxims (Cooperative Principle) from conversational design frameworks, ensuring the chatbot’s responses are clear, relevant, and natural, which makes the interaction more engaging and easy to follow.
USER PAIN POINT
"I don’t always know what to type, and I’m not sure what the chatbot expects."
SOLUTION
The chatbot offers pre-filled answer suggestions to guide users, helping them quickly choose an option or see examples of what they can say. This reduces hesitation and confusion.
This follows Mixed-Initiative Interaction from Google’s Conversational Design Principles, allowing both the user and the chatbot to drive the conversation, making it smoother and more collaborative.
USER PAIN POINT
"Most chatbots restrict me from typing my own answers."
SOLUTION
Users can type custom answers instead of only selecting from pre-set options, giving them the freedom to share unique preferences or details.
This follows User Control & Flexibility, a key principle from Nielsen’s Usability Heuristics, ensuring the chatbot adapts to different user styles and needs, not the other way around.
USER PAIN POINT
"I want to update my responses if my plans change, but the chatbot locks my answers."
SOLUTION
Users can go back and edit their previous responses, so they don’t have to start over if they want to change their plans, making the chatbot more adaptable.
This applies Error Recovery & Correction, from Microsoft’s Conversational UX Principles, which encourages letting users fix mistakes easily, reducing frustration and improving flexibility.
How I arrived at this chatbot design
I focused on the interface design of the chatbot, and building a simulation prototype to visualize its use-cases. The process of constructing this prototype involved two main stages– designing the chatbot persona and outlining the conversation flow.
CONSTRUCTION OF THE CHATBOT
Step 1: Defining the chatbot persona
The chatbot’s persona is “a friendly, well-traveled companion who helps you plan your perfect trip, suggests ideas, shares fun facts, and keeps the conversation casual and helpful — like chatting with a travel-savvy friend.”
Step 2: Outlining the conversation flow
For this stage, I created a simple, linear dialogue flow with the objective of building an itinerary. I started by trying the conversation as spoken conversation, to ensure it felt natural and user-friendly.
Here is the draft of the dialogue flow I used to build the final chatbot prototype–
Here is the draft of the dialogue flow I used to build the final chatbot prototype–
Curious about the product? Try the prototype now!
View PrototypeIMPACT
The flexibility, and ease of use of EnRoute's itinerary planner were refined through multiple iterations. And the results? The users loved it! They appreciated its natural tone and ability to personalize itineraries effortlessly.
88.4 on the System Usability Scale*
Participants agreed that the application was easy to use.
*A standardized questionnaire used as a measure of usability
75% Perceived Adoption Rate*
Participants strongly agreed they would use the app frequently.
*How likely users are to keep using a product after trying it
FUTURE SCOPE
Moving Beyond Simulation & Developing a Working MVP
The immediate next step is to explore data sources for the chatbot to to understand where it could gather data from. I also want to evaluate chatbot builder platforms like Dialogflow or Landbot to simplify development & deployment. This will help transition the chatbot from a design prototype to a functional system.
Making Error Handling
Smoother
Smoother
To make the chatbot easier to use, I want to improve how it handles errors. If a user types something unexpected, the chatbot could suggest options or guide them back on track instead of just showing an error. The goal is to make sure users always know what to do next.
REFLECTIONS & LEARNINGS
Learning to Collaborate Better
Working on the EnRoute project taught me how to collaborate more effectively — not just within my design team, but also with stakeholders. Balancing feedback, aligning on goals, and adapting to different perspectives helped me build stronger communication and teamwork skills.
Working in an Early-Stage Startup Environment
This project gave me a glimpse into how early-stage startups work, where every team member wears multiple hats. From taking ownership of design decisions to pitching ideas, presenting progress, and helping manage the overall project, I learned how to step up, stay flexible, and think beyond just design.
The team behind this project
