AI Travel Companion
A comprehensive AI-powered travel planning assistant that transforms how users plan trips. Users input their chosen city and number of days, and the app generates customized day-by-day itineraries.
Role
Full-stack engineer
Team
With Anas Shaik and Rahul Sumbly (AI Engineers at Deriv)
Company/Organization
BITS (University Project)
The Problem
Travelers had to use multiple apps and websites for research, planning, and navigation.
Creating day-by-day itineraries required hours of manual research and organization.
Route optimization between multiple destinations was time-consuming and often suboptimal.
No single tool provided personalized recommendations based on trip duration.
Manual planning often resulted in inefficient routes and missed key attractions.
The Solution
The AI Travel Companion applies AI techniques including natural language processing and search algorithms to transform travel planning.
Users simply input their destination city and number of days, and the system handles the rest.
Day-by-Day Itineraries: The app provides a unique itinerary for each day, ensuring a well-paced and enjoyable travel experience.
Each day has a start location, activity list, and end location, with the system optimizing within each day for efficiency.
AI-Powered Suggestions: Utilizing an LLM, the app generates a list of must-visit places based on the number of days available.
This ensures travelers make the most of their time and don't miss key attractions.
The LLM handles understanding user preferences, suggesting destinations, and creating narratives around the trip.
Optimized Routing: Employing advanced pathfinding algorithms, the app calculates the shortest route for each day's itinerary, covering all...
This minimizes travel time and maximizes exploration.
Visualized on Maps: The optimized routes are visually presented on interactive maps, allowing users to easily follow their daily journeys and...
Google Maps Integration: A seamless Google Maps navigation link is provided for each day's itinerary, making on-the-ground navigation effortless with...
Design Decisions
Used LLM for trip suggestions instead of a recommendation engine. LLMs can generate reasonable plans with zero training data, unlike recommendation...
Chose structured JSON output over free-text responses for reliable parsing and consistent UI display of itineraries.
Integrated Google Maps API for real-time data (distances, traffic, place details) to augment LLM responses which lack current information.
Implemented advanced pathfinding algorithms to calculate shortest routes covering all recommended places each day.
Designed day-by-day state machine where each day has start location, activities, and end location with optimization within each day.
Used client-side state management (React Context) with localStorage for single-session trip planning.
Tradeoffs & Constraints
Time constraint: Built MVP in 4 weeks, focusing on core itinerary generation and route optimization.
API cost: LLM and Google Maps API calls controlled through caching and limited retries.
Data constraint: LLM knowledge cutoff means relying on Google Maps for current place information.
Future scope: Customizable itineraries (add/remove places), food stop suggestions, and rest break recommendations.
Would improve: Add collaborative trip planning for multiple users, budget tracking, and booking integration.
Outcome & Impact
Transformed travel planning from hours of research to minutes of input.
AI-powered suggestions ensure travelers don't miss key attractions based on trip duration.
Optimized routing minimizes travel time and maximizes daily exploration efficiency.
Interactive map visualization helps users understand and follow their daily journeys.
Google Maps integration provides seamless on-the-ground navigation with turn-by-turn directions.
Personalized day-by-day itineraries ensure well-paced and enjoyable travel experiences.
Tech Stack
Next.js for application framework and API routes
React for interactive UI and state management
LLM API for AI-powered place suggestions and itinerary generation
Google Maps API for navigation, place data, and distance calculations
Advanced pathfinding algorithms for route optimization
Natural Language Processing (NLP) for understanding user inputs