Wanderly: An AI-Powered Full-Stack Travel Planning Web Application Using Large Language Models
TL;DR
Wanderly is presented, a full-stack AI-powered travel planning web application that leverages the LLaMA 3.3 70B model via Groq API to generate personalized, structured travel itineraries from user-specified preferences encompassing budget constraints, travel interests, trip duration, traveler count, and destination.
Abstract
The rapid advancement of large language models (LLMs) has created transformative opportunities for intelligent consumer applications. This paper presents Wanderly, a full-stack AI-powered travel planning web application that leverages the LLaMA 3.3 70B model via Groq API to generate personalized, structured travel itineraries from user-specified preferences encompassing budget constraints, travel interests, trip duration, traveler count, and destination. The system architecture integrates a React 18 + Vite frontend with a Node.js/Express backend, Supabase for PostgreSQL-based authentication and data persistence, and a split-deployment pipeline across Vercel and Render. Wanderly addresses the fragmented nature of conventional travel planning by delivering end-to-end itinerary generation, categorical budget breakdowns, curated activity recommendations, contextual insider guidance, destination photo collages, and aggregated booking resource links within a unified platform. Systematic evaluation across 50 test scenarios demonstrates 100% structurally valid JSON responses, 98% destination adherence accuracy, and average AI inference times of 11.3 seconds. This work details system architecture, prompt engineering methodology, multi-interest UX paradigm, deployment pipeline, and a future enhancement roadmap for real-time API integration.